| blob | e91641153003d229e3af54744eb92f12256dbecf |
1 2010-06-16 Andreas J. Koenig <andk@cpan.org>
3 * rrr-init seems to be sssslllloooowwww. At 3:49 I started it for the
4 video tree. After twelve minutes the file has reached a size of 500k. It
5 is locking and rewriting the YAML after every file. Wasn't there an
6 unlocked variant? For "init" one would not expect locked correctness and
7 a rewrite after every item. Given that this tends to become slower with
8 every file it is really unacceptable.
10 Reminds me of rrr-dirtyupdate (the former name of rrr-update). It seems
11 the code was not dirty enough and that was the reason why it was so
12 slow. And rrr-update does not have a --dirty switch, so it looks like we
13 have no way to force dirty and fast operation yet but we need it for
14 init.
16 2010-06-15 Andreas J. Koenig <andk@cpan.org>
18 * remove the wild alpha alerts and say, it is alpha but parts of it are
19 already working very well or so.
21 2009-12-10 Andreas J. Koenig <andk@cpan.org>
23 * Prominent request for statusfile to safe memory consumption. I'd like
24 to work backwards: get a verification program that verifies how well in
25 sync two nodes are without using frmr itself. Pretend that the forking
26 agent already works and can be interrupted with ^C. Then let it die at
27 random and restart at random and see the memory consumption and the
28 failure rate over time. Compare it with the curve we have now.
30 Of course we also invented the status file as a sharp debugging tool.
31 I'd like to see it in action. See also: F:R:M:Recent::thaw().
33 2009-06-22 Andreas J. Koenig <andk@cpan.org>
35 * Todo: rrr-init to generate some index files so that rrr-server can be
36 started.
38 2009-05-02 Andreas J. Koenig <andk@cpan.org>
40 * DONE: rename rrr-dirtyupdate to rrr-update and add an option --dirty
41 so we can use it for normal update with a single file. And add a cronjob
42 that runs rrr-fsck so we get a reminder when something is not working
43 as it should.
45 2009-04-27 Andreas J. Koenig <andk@cpan.org>
47 * Todo: I still see more link_stat errors than expected. I come back to
48 the former suspicion that we do not do necessary bookkeeping on deleted
49 files so that when we reach the "new" event on a meanwhile deleted file
50 we try to mirror it although we could know better. It's just noise, no
51 harm but noise is irritating.
53 * Todo: more experimenting with runstatusfile to get it official
55 * Todo: make sure $ENV{LANG}="C" when we call rsync because we will
56 parse the error.
58 rsync: link_stat "/authors/id/T/TH/THINC/DateTime-Format-Flexible-0.02.tar.gz" (in PAUSE) failed: No such file or directory
60 rsync: link_stat "/authors/id/A/AG/AGENT/SSH-Batch-0.001.meta" (in PAUSE) failed: No such file or directory (2)
62 And that ignore_link_stat_error is propagated in sparse_clone.
64 DONE
66 For the record: the bug was that ignore_link_stat_error was lost during
67 sparse_clone. And then the second bug was that it did not default to
68 true. Both are now fixed.
70 2009-04-25 Andreas J. Koenig <andk@cpan.org>
72 * Todo: fill rrr program with life
74 2009-04-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
76 * Minor bug: I see complaints about files not existing like
77 /authors/id/G/GW/GWADEJ/SVG-Sparkline-0.2.0.meta or
78 /authors/id/R/RC/RCAPUTO/POE-1.004.readme
80 In the RECENT files these are delete events but of course also new
81 events in older recent files.
83 I think we do not properly remember deletes when we skip-deletes and so
84 do not filter them out when we later see the "new" event.
86 Yes, but that what ignore_link_stat_error is for, we know about this
87 race condition. We must make ignore_link_stat_error default to true.
89 DONE
91 * bug in Done.pm failing to merge two fields into one when a third is
92 present. FIXED and accompanied with new test.
94 2009-04-24 Andreas J. Koenig <andk@cpan.org>
96 * Todo: keep the index clean and run some kind of nondestructive fsck 4
97 times a day.
99 * who is our backbone?
101 http://cpan.cpantesters.org/authors/02STAMP # barbie
102 http://cpan.solfo.com/authors/02STAMP # abh
103 (http|ftp)://theoryx5.uwinnipeg.ca/pub/CPAN/ # rkobes
105 2009-04-17 Andreas J. Koenig <andk@cpan.org>
107 * the changelog helper I included in the Makefile release memo does not
108 show tags and is much less useful than I thought. gitk is probably much
109 more convenient.
111 2009-04-16 Andreas J. Koenig <andk@cpan.org>
113 * ABH writes:
115 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
116 #### Ydr_.yaml ... DONE
117 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
118 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
119 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
120 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
121 #### gone already at cpan-pause.pl line 0
123 Running without skip-deletes now but cannot reproduce.
125 2009-04-15 Andreas J. Koenig <andk@cpan.org>
127 * consider state directory such we can restart after a ^C or come again
128 from a cronjob.
130 Reuse _runstatusfile: write atomically, see how we can use the dump to
131 start running again. Do not leak and remove globs before writing. Do not
132 only dump one rf, dump the whole complex.
134 * consider the effect of ^C. Is it safe?
136 * want to use Perl::Version to up the versions on every release. Only if
137 things have changed, of course. Does this actually work?
139 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
140 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
142 Must do the bookkeeping myself when this should happen, maybe based on git?
144 I think I prefer the same version number for all pm files. But one also
145 needs a routine for setting them. Ahh:
147 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
149 NEEDSMOREWORK
151 * why indexer wrong?
153 cpan[2]> m /Mirror::Recent/
154 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
155 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
156 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
157 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
158 4 items found
160 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
162 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
164 I just tried MM 6.5102 and it again produces only a provides for Recent,
165 not for the other modules.
167 My own fault. FIXED in the Makefile.PL.
169 2009-04-13 Andreas J. Koenig <andk@cpan.org>
171 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
173 * is it true (as stated at
174 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
175 that rsync can lead to errors when upstream changes before client sync
176 has completed? Can we see the error when we run this on upstream:
178 % perl -e '
179 use Time::HiRes qw(time);
180 while (){
181 open my $fh, ">", "changingfile.txt.new" or die;
182 print $fh (time."\n") x 1000000;
183 close $fh;
184 rename "changingfile.txt.new", "changingfile.txt" or die;
185 }
186 '
188 And on the receiving end:
190 % while true; do
191 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
192 sleep 1;
193 done
195 I cannot get it to fail. Apparently it is sufficient when upstream
196 always writes atomically (which of course is mandatory)?
198 SENT to mailing list Apr 28.
200 * from ABH:
202 > https://fedorahosted.org/InstantMirror/
203 > https://www.redhat.com/mailman/listinfo/instantmirror-list
204 > irc.freenode.net channel #instantmirror
206 A cool name for a project. Inspires me to write a few sentences about
207 bittorrent's role in the grand picture.
209 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
210 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
212 2009-04-12 Andreas J. Koenig <andk@cpan.org>
214 * Interesting last minute bug during real download testing: the output
215 isn't as pretty anymore, it seems that more work is being done than
216 needed.
218 Introducing a highlevel _alluptodate method helped a lot in debugging
219 this.
221 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
222 different on 1h file than on the other files. This lead to frequent
223 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
224 issue.
226 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
228 * Todo: try if we can be faster with a native float package. I'm really
229 glad to have a machine exposing perl floating point bugs, it forced me
230 to retract this item from the changes file: "several speedups in the
231 fakebigfloat code" before I get cpan testers fails.
233 2009-04-11 Andreas J. Koenig <andk@cpan.org>
235 * Bug: when the dirtymark changes upstream, then the downstream server
236 notices it immediately and mirrors the "1h" file but then leaves the
237 rmirror loop and continues with "6h" on the next round through. This
238 behaviour goes on until it reaches the Z file. Because we eat the Z file
239 piecemeal we leave the loop after a while but after that everybody seems
240 to have forgotten that there is some work left to be done. Bad, bad,
241 bad.
243 ETOOCONFUSING better name?
245 Recentfile::get_remote_recentfile_as_tempfile OK
246 Recentfile::resolve_recentfilename split_rfilename DONE
248 Recent::_principal_recentfile_object _principal_recentfile_fromremote
249 Recent::principal_recentfile OK
250 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
252 some void, some not, some not void but called in void context.
254 2009-04-10 Andreas J. Koenig <andk@cpan.org>
256 * Bug? When the dirtymark changes then the second tier hosts quickly
257 reset their DONE state and restart mirroring. But then they mirror
258 potentially outdated index files with inconsistent dirtymark because the
259 first tier box may be a bit behind on the large recentfiles.
261 This means we need a brake when iterating over the recentfiles that
262 refuses to use a recentfile with the wrong dirtymark.
264 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
266 * install the bin/* files and talk about rrr-overview somewhere.
268 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
269 created outside the target tree.
271 * Barbie asks for better logging capabilities
273 2009-03-29 Andreas J. Koenig <andk@cpan.org>
275 * known bug: cannot configure the Recentfile objects to keep delete
276 events. How to call the accessor for that? keep_delete_objects_forever?
278 Sounds acceptable. FIXED (but untested).
280 * known bug: keeps temporary index files lying around.
282 2009-03-24 Andreas J. Koenig <andk@cpan.org>
284 * Some equivalent for
286 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
288 ? DONE
290 2009-03-22 Andreas J. Koenig <andk@cpan.org>
292 * k81 is again client of k75 with different parameters.
294 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
295 about it all the time and nobody reacts accordingly; (2) lots of temp
296 files get created and not removed; culprit the new call to
297 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
298 caller has to remove the tempfile after use.
300 Need the drawing board.
302 PARTLY FIXED: retracting the idea to call
303 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
304 and unseed such that uptodate now means "we have mirrored this
305 recentfile's bunch of files" and seeded means "somebody believes the
306 index file for this rf needs be refreshed". This seems to work now.
308 Still collecting temp files that nobody cares to depollute.
310 2009-03-21 Andreas J. Koenig <andk@cpan.org>
312 * Bug?: should it be harder than it is atm to set the timestamp to the
313 future? YES, FIXED
315 * bug with native integers:
317 -
318 epoch: 997872011
319 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
320 type: new
321 -
322 epoch: 1195248431
323 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
324 type: new
326 Native integer broke when native math was turned off. FIXED
328 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
329 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
330 in this area.
332 Yes, records out of order:
334 447003 -
335 447004 epoch: 995885533
336 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
337 447006 type: new
338 447007 -
339 447008 epoch: 995890358
340 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
341 447010 type: new
342 447011 -
343 447012 epoch: 995892221
344 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
345 447014 type: new
347 FIXED with sanity check and later with the integer fix.
349 * Bug: want the index files in a .recent directory
351 * Bug: lots of dot files are not deleted in time
353 * possible test case: can a delete change the timestamp? This would
354 probably break the order of events.
356 2009-03-20 Andreas J. Koenig <andk@cpan.org>
358 * 1233701831.34486 what's so special about this number/string? It
359 belongs to
360 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
361 atm lives in Y,Q, and Z.
363 It is the first entry after id/--skip-locking which has timestamp
364 1234164228.11325 which represents a file that doesn't exist anymore.
366 * another bug:
368 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
369 _bigfloatcmp called with l[1237505213.21133]r[UNDEF]: but both must be defined at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
370 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
371 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
372 File::Rsync::Mirror::Recentfile::Done::covered('File::Rsync::Mirror::Recentfile::Done=HASH(0x8857fb4)', 1237505213.21133, 0.123456789) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile.pm line 2041
373 File::Rsync::Mirror::Recentfile::uptodate('File::Rsync::Mirror::Recentfile=HASH(0x8533a2c)') called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recent.pm line 536
374 File::Rsync::Mirror::Recent::rmirror('File::Rsync::Mirror::Recent=HASH(0x82ef3d0)', 'skip-deletes', 1) called at /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl line 27
375 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
376 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
377 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
378 File::Rsync::Mirror::Recentfile::Done::covered('File::Rsync::Mirror::Recentfile::Done=HASH(0x8857fb4)', 1237505213.21133, 0.123456789) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile.pm line 2041
379 File::Rsync::Mirror::Recentfile::uptodate('File::Rsync::Mirror::Recentfile=HASH(0x8533a2c)') called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recent.pm line 536
380 File::Rsync::Mirror::Recent::rmirror('File::Rsync::Mirror::Recent=HASH(0x82ef3d0)', 'skip-deletes', 1) called at /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl line 27
382 FIXED, it was the "--skip-locking" file where manual intervention was participating
384 * bug on the mirroring slave: when the dirtymark gets increased we
385 probably do not reset the done intervals. The mirrorer stays within
386 tight bounds where it tries to sync with upstream and never seems to
387 finish. In the debugging state file I see lots of identical intervals
388 that do not get collapsed. When I restart the mirrorer it dies with:
390 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
391 _bigfloatcmp called with l[1237400817.94363]r[UNDEF]: but both must be defined at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
392 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
393 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
394 File::Rsync::Mirror::Recentfile::Done::_register_one('File::Rsync::Mirror::Recentfile::Done=HASH(0x84c6af8)', 'HASH(0xb693f2dc)') called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 200
395 File::Rsync::Mirror::Recentfile::Done::register('File::Rsync::Mirror::Recentfile::Done=HASH(0x84c6af8)', 'ARRAY(0x8c54bfc)', 'ARRAY(0xb67e618c)') called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile.pm line 1044
396 File::Rsync::Mirror::Recentfile::_mirror_item('File::Rsync::Mirror::Recentfile=HASH(0x84abcb0)', 227, 'ARRAY(0x8c54bfc)', 33110, 'File::Rsync::Mirror::Recentfile::Done=HASH(0x84c6af8)', 'HASH(0x84abd8c)', 'ARRAY(0x839cb64)', 'HASH(0x839c95c)', 'HASH(0xb6a30f0c)', ...) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile.pm line 992
397 File::Rsync::Mirror::Recentfile::mirror('File::Rsync::Mirror::Recentfile=HASH(0x84abcb0)', 'piecemeal', 1, 'skip-deletes', 1) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recent.pm line 564
398 File::Rsync::Mirror::Recent::_rmirror_mirror('File::Rsync::Mirror::Recent=HASH(0x84ab6d4)', 7, 'HASH(0x8499488)') called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recent.pm line 532
399 File::Rsync::Mirror::Recent::rmirror('File::Rsync::Mirror::Recent=HASH(0x84ab6d4)', 'skip-deletes', 1) called at /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl line 27
401 and debugging stands at
403 At: before
404 Brfinterval: Z
405 Ci: 227
406 Dre+1:
407 epoch: 1237400802.5789
408 path: id/J/JO/JOHND/CHECKSUMS
409 type: new
410 Dre-0:
411 epoch: 1237400807.97514
412 path: id/M/MI/MIYAGAWA/CHECKSUMS
413 type: new
414 Dre-1:
415 epoch: 1237400817.94363
416 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
417 type: new
418 Eintervals:
419 -
420 - 900644040
421 - 900644040
422 - []
424 and it is reproducable.
426 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
427 pause has a fresh one.
429 FIXED, it was the third anded term in each of the ifs in the IV block in
430 _register_one: with that we make sure that we do not stamp on valuable
431 interval data.
433 2009-03-17 Andreas J. Koenig <andk@cpan.org>
435 * done: verified the existence of the floating point bug in bleadperl
436 and verified that switching from YAML::Syck to YAML::XS does not resolve
437 it.
439 BTW, the switch was doable with
441 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
443 and should be considered as a separate TODO
445 * todo: integrate a dirty update with two aggregate calls before
446 unlocking for frictionless dirtying DONE
448 * todo: start the second rsync daemon on pause DONE
450 * todo: move index files to .recent: this cannot simply be done by
451 setting filenameroot to .recent/RECENT. Other parts of the modules rely
452 on the fact that dirname(recentfile) is the root of the mirrored tree.
454 2009-03-16 Andreas J. Koenig <andk@cpan.org>
456 * What was the resolution of the mirror.pl delete hook bug? Do we call
457 the delete hook when pause removes a file from MUIR?
459 * Today on pause: Updating 2a13fba..29f284d and installing it for
460 /usr/local/perl-5.10.0{,-RC2}
462 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
463 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
465 2009-03-15 Andreas J. Koenig <andk@cpan.org>
467 * currently recent_events has the side effect of setting dirtymark
468 because it forces a read on the file. That should be transparent, so
469 that the dirtymark call always forces a cache-able(?) read.
471 * The bug below is -- after a lot of trying -- not reproducible on a
472 small script, only in the large test script. The closest to the output
473 below was:
475 #!perl
476 use strict;
477 use Devel::Peek;
478 my $x = "01237123229.8814";
479 my($l,$r);
480 for ($l,$r) {
481 $_ = "x"x34;
482 }
483 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
484 $r = int $r;
485 $l = "1237123231.22458";
486 $r = "1237123231.22458";
487 1 if $l/1.1;
488 Devel::Peek::Dump $l;
489 Devel::Peek::Dump $r;
490 Devel::Peek::Dump $x = $l <=> $r;
491 die "BROKE" if $x;
492 __END__
494 The checked in state at c404a85 fails the test with my
495 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
496 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
497 in the test script always consistent.
499 * Todo: write a test that inserts a second dirty file with an already
500 existing timestamp. DONE
502 * Bug in perl 5.10 on my 64bit box:
504 DB<98> Devel::Peek::Dump $l
505 SV = PVMG(0x19e0450) at 0x142a550
506 REFCNT = 2
507 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
508 IV = 0
509 NV = 1237123231.22458
510 PV = 0x194ce70 "1237123231.22458"\0
511 CUR = 16
512 LEN = 40
514 DB<99> Devel::Peek::Dump $r
515 SV = PVMG(0x19e0240) at 0x142a3e8
516 REFCNT = 2
517 FLAGS = (PADMY,POK,pPOK)
518 IV = 1237123229
519 NV = 1237123229.8814
520 PV = 0x19ff900 "1237123231.22458"\0
521 CUR = 16
522 LEN = 40
524 DB<100> Devel::Peek::Dump $l <=> $r
525 SV = IV(0x19ea6e8) at 0x19ea6f0
526 REFCNT = 1
527 FLAGS = (PADTMP,IOK,pIOK)
528 IV = -1
530 DB<101> Devel::Peek::Dump $l
531 SV = PVMG(0x19e0450) at 0x142a550
532 REFCNT = 2
533 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
534 IV = 0
535 NV = 1237123231.22458
536 PV = 0x194ce70 "1237123231.22458"\0
537 CUR = 16
538 LEN = 40
540 DB<102> Devel::Peek::Dump $r
541 SV = PVMG(0x19e0240) at 0x142a3e8
542 REFCNT = 2
543 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
544 IV = 1237123231
545 NV = 1237123231.22458
546 PV = 0x19ff900 "1237123231.22458"\0
547 CUR = 16
548 LEN = 40
550 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
552 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
554 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
556 * $obj->merge ($other) needs to learn about equal epoch which may happen
557 since dirty_epoch intruded.
559 * Wontfix anytime soon: I think we currently do not support mkdir. Only
560 files!
562 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
564 * Todo: continue working on update(...,$dirty_epoch). It must be
565 followed by a fast_aggregate! DONE
567 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
569 * maybe we need a closest_entry or fitting_interval or something like
570 that. We want to merge an event into the middle of some recentfile.
571 First we do not know which file, then we do not know where to lock,
572 where to enter the new item, when and where to correct the dirtymark.
574 So my thought is we should first find which file.
576 Another part of my brain answers: what would happen if we would enter
577 the new file into the smallest file just like an ordinary new event,
578 just as an old event?
580 (1) we would write a duplicate timestamp? No, this would be easy to
581 avoid
583 (2) we would make the file large quickly? Yes, but so what? We are
584 changing the dirtymark, so are willing to disturb the downstream hosts.
586 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
588 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
590 leak!
592 https://rt.cpan.org/Ticket/Display.html?id=41199
594 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
596 Strangest output being HAKANARDO who managed to upload
598 Here is a better oneliner that includes also the first line of each
599 finding:
601 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$F[-1]=~s/\s+\z//; push @{$Seen{$F[-1]}||=[]},$_; END {for my $k (keys %Seen){ delete $Seen{$k} if @{$Seen{$k}}==1; } print YAML::Syck::Dump(\%Seen)}'
603 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
605 * memory leak in the syncher? It currently weighs 100M.
607 Update 2008-11-02:
609 root 10705 1.0 4.9 80192 76596 pts/32 S+ Nov02 24:05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
612 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
614 * lookup by epoch and by path and use this ability on the pause to never
615 again register a file twice that doesn't need it. Let's call it
616 contains().
618 * after the dirtymark is done: fill up recentfiles with fake (historic)
619 entries; fill up with individual corrections; algorithm maybe to be done
620 with bigfloat so that we can always place something in the middle
621 between two entries. Before we must switch to bigfloat we could try to
622 use Data::Float::nextup to get the.
624 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
625 directly.
627 * dirtymark now lives in Recentfile, needs to be used in rmirror.
629 * find out why the downloader died after a couple of hours without a net
630 connection. Write a test that survives the not-existence of the other
631 end forever.
633 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
635 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
636 it makes less sense with tight coupling for secondary files. But asking
637 the server what the current epoch is might be cheaper on HTTP than on
638 rsync. (Needs to be evaluated)
640 * remove the 0.00 from the verbose overview in the Merged column in the
641 Z row. DONE
643 * write tests that expose the problems of the last few days: cascading
644 client/server roles, tight coupling for secondary RFs, deletes after
645 copies.
647 * Some day we might want to have policy options for the slave:
648 tight/loose/no coupling with upstream for secondary RFs. tight is what
649 we have now. loose would wait until a gap occurs that can be closed.
651 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
653 * revisit all $rfs->[$i+1] places if they now make sense still
655 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
657 * another bug is the fact that the mirror command deletes files before
658 it unhides the index file, thus confusing downstream slaves. We must not
659 delete before unhiding and must delete after unhiding. FIXED.
661 * new complication about the slave that is playing a server role.
662 Currently we mirror from newest to oldest with a hidden temporary file
663 as index. And when one file is finished, we unhide the index file.
664 Imagine the cascading server/slave is dead for a day. It then starts
665 mirroring again with the freshest thing and unhides the freshest index
666 file when it has worked through it. In that moment it exposes a time
667 hole. Because it now works on the second recentfile which is still
668 hidden.
670 We currently do nothing special to converge after such a drop out. At
671 least not intentionally and robustly and thought through.
673 The algorithm we use to seed the next file needs quite a lot of more
674 robustness than it currently has. Something to do with looking at the
675 merged element of the next rf and when it has dropped off, we seed
676 immediately. And if it ramains dropped off, we seed again, of course.
678 Nope, looking from smaller to larger RFS we look at the merged element
679 of this RF and at the minmax/max element of the next RF. If that
680 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
681 seeding it.
683 And we need a public accessor seed and unseed or seeded. But not the mix
684 of public and private stuff that then is used behind the back.
686 And then the secondary* stuff must go.
688 And we must understand what the impact is on the DONE system. Can it go
689 unnoticed that there was a hole? And could the DONE system have decided
690 the hole is covered? This should be testable with three directories where
691 the middle stops working for a while. Done->merge is suspicious, we must
692 stop it from merging non-conflatable neighbors due to broken continuity.
694 FIXED
696 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
698 * Slaven suggests to have the current epoch or the whole current
699 recentfile available from the HTTP server and take it away with
700 keepalive. This direction goes the granularity down to subseconds.
702 We might want to rewrite everything to factor out transport and allow
703 the whole thing to run via HTTP.
705 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
707 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
708 symlink?
710 * smoker on k81 fetching from k75 to verify cascading works. See
711 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
713 * maybe the loop should wait for CHECKSUMS file after every upload. And
714 CPAN.pm needs to deal with timestamps in the future.
716 * do not forget the dirtymark!
718 Text: have a new flag on recentfiles with the meaning: if this
719 changes, you're required to run a full rsync over all the files. The
720 reason why we set it would probably be: some foul happened. we injected
721 files in arbitrary places or didn't inject them although they changed.
722 The content of the flag? Timestamp? The relation between the
723 recentfiles would have to be inheritance from the principal, because any
724 out of band changes would soon later propagate to the next recentfile.
726 By upping the flag often one can easily ruin the slaves.
728 last out of band change? dirtymark?
730 Anyway, this implies that we read a potentially existing recentfile
731 before we write one.
733 And it implies that we have an eventloop that keeps us busy in 2-3
734 cycles, one for current stuff (tight loop) and one for the recentfiles
735 (cascade when principal has changed), one for the old stuff after a
736 dirtymark change.
738 And it implies that the out-of-band change in any of the recentfiles
739 must have a lock on the principal file and there is the place to set the
740 dirtymark.
742 * start a FAQ, especially quick start guide questions. Also to aid those
743 problematic areas where we have no good solution, like the "links"
744 option to rsync.
746 * wish feedback when we are slow.
748 * reduce mccabe
750 * Remove a few DEBUG statements.
752 * The multiple-rrr way of doing things needs a new option to rmirror,
753 like piecemeal or so. Not urgent because after the first pass through,
754 things run smoothely. It's only ugly during the first pass.
756 * I have the suspicion that the code is broken that decides if the
757 neighboring RF needs to be seeded. I fear when too much time has gone
758 between two calls (in our case more than one hour), it would not seed
759 the neighbor. Of course this will never be noticed, so we need a good
760 test for it.
762 * local/localroot confusion: I currently pass both options but one must
763 do.
765 * accounts for early birds on PAUSE rsync daemon.
767 * hardcoded 20 seconds
769 * who mirrors the index? DOING now.
771 * which CPAN mirrors offer rsync?
773 * visit all XXX, visit all _float places
775 * rename the pathdb stuff, it's too confusing. No idea how.
777 * rrr-inotify, backpan, rrr-register
779 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
781 * current bugs: the pathdb seems to get no reset, the seeding of the
782 secondaryttl stuff seems not to have an effect. Have helped myself with
783 a rand(10), need to fix this back. So not checked in. Does the rand
784 thing even help?
786 The rand thing helps. The secondaryttl stuff was in the wrong line,
787 fixed now.
789 The pathdb stuff was because I called either _pathdb or __pathdb on the
790 wrong object. FIXED now.
792 * It's not so beautiful if we never fetch the recentfiles that are not
793 the principal, even if this is correct behaviour. We really do not need
794 them after we have fetched the whole content.
796 OK, we want a switch for that: secondaryttl DONE
798 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
800 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
801 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
802 it lets 40 follow. FIXED
804 * See that the long running process really only updates the principal
805 file unless it has missed a timespan during which something happened. If
806 nothing happened, it must notice even when it misses the timespan. DONE
808 * we must throw away the pathdb when we have reached the end of Z. From
809 that moment we can have a very small pathdb because the only reason for
810 a pathdb is that we know to ignore old records in old files. We won't
811 need this pathdb again before the next full pass over the data is
812 necessary and then we will rebuild it as we go along. DONE
814 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
816 * I think, Done::register_one is doing wrong in that it does not
817 conflate neighboring pieces. The covered() method cannot do this because
818 it has no recent_events array at hand. But register_one has it and could
819 do it and for some reason misses to do it (sometimes).
821 This means that the three tests I just wrote can probably not survive
822 because they test with an already broken Done structure.
824 The art now is to detect how it happens, then to reproduce, then write a
825 test, then fix it.
827 So from the logfile this is what happens: we have a good interval with
828 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
829 top of it. Locally we now mirror F2 and open a new done interval for it.
830 Then we mirror F1 but this time with the timestamp T1b. And when we then
831 try to close the gap, we do not find T1 but instead something older. We
832 should gladly accept this older piece and this would fix this bug.
834 FIXED
836 * bug to fix: when the 1h file changes while rmirror is running, we do
837 correctly sync the new files but never switch to the 6h file but rather
838 stay in a rather quick loop that fetches the 1h file again and again.
840 Is it possible that we initialize a new object? Or does
841 get_remote_recentfile_as_tempfile overwrite something in myself?
843 Want a new option: _runstatusfile => $file which frequently dumps the
844 state of all recentfiles to a file.
846 FIXED
848 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
850 * Todo: now teach update to verify the timestamp is about to write
851 against the previous and use _increase_a_bit if it doesn't comply with
852 strict monotony. DONE
854 * The problem of rounding. So far perl's default precision was
855 sufficient. One day it won't be. FakeFloat has an easy job when it is
856 only reading and other machines have written correctly. But when we want
857 to write a floating point number that is a bit larger than the other
858 one, then we need our own idea of precision.
860 Slaven said: just append a "1". This might be going towards the end of
861 usability too quickly. I'd like something that actually uses the decimal
862 system. Well, appending a 1 also does this but...
864 E.g. we have 1.0. nextup on this architecture is starting with
865 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
866 taken the 1.0000000000000003 and the next user comes and the time tells
867 him 1.0 again. He has to beat my number without stepping over the
868 nextup. This is much less space than I had when I chose 1,2,3.
870 What is also irritating is that nextup is architecture dependent. The
871 128 bit guy must choose very long numbers to fit in between whereas the
872 other one with 16 bit uses larger steps. But then the algorithm is the
873 same for both, so that would be a nice thing.
875 I see two situation where we need this. One is when Time::HiRes returns
876 us a value that is <= the last entry in our recentfile. In this case
877 (let's call it the end-case) we must fill the region between that number
878 and the next higher native floating point number. The other is when we
879 inject an old file into an old recentfile (we would then also set a new
880 dirtymark). We find the integer value already taken and need a slightly
881 different one (let's call it the middle-case). The difference between
882 the two situations is that the next user will want to find something
883 higher than my number in the end-case and something lower than my number
884 in the middle case.
886 So I suggest we give the function both a value and an upper bound and it
887 calculates us a primitive middle. The upper bound in the middle-case is
888 the next integer. The upper bound on the end-case is the nextup floating
889 point number. But the latter poses another problem: if we have occupied
890 the middle m between x and nextup(x), then the nextup(m) will probably
891 not be the same as nextup(x) because some rounding will take place
892 before the nextup is calculated and when the rounding reaches the
893 nextup(x), we will end up at nextup(nextup(x)).
895 So we really need to consider the nextup and the nextdown from there and
896 then the middle and that's the number we may approach asymptotically.
897 Ugly. But DONE.
899 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
901 * consider deprecating the use of RECENT.recent as a symlink. It turns
902 out to need extra hoops with the rsync options and just isn't worth it.
903 Or maybe these extra hoops are needed anyway for the rest of the tree?
904 Nope, can't be the case because not all filesystems support symlinks.
906 But before doing the large step, I'll deprecate the call of
907 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
908 only introduced to resolve RECENT.recent and complicates the routine far
909 beyond what it deserves.
911 DONE. Won't deprecate RECENT.recent, just moved its handling to the
912 supervisor.
914 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
916 * I think it's a bug that the rsync_option links must be set to true in
917 order to support RECENT.recent and that nobody cares to set it
918 automatically. Similar for ignore_link_stat_errors.
920 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
922 * Todo: find all todos together and make a plan what is missing for a
923 release.
925 - verifytree or something like that. fsck maybe.
927 - rersyncrecent, the script itself? What it do?
929 - a way to only mirror the recentfiles without mirroring the whole
930 remote system such that people can decide to mirror only partially see
931 also 2008-08-30. .shadow-xxx directory? this also needed for a
932 filesystem that is still incomplete and might need the mirrorfiles for
933 lookup(?)
935 - long living objects that mirror again and again. Inject something
936 into ta, see how it goes over to tb.
938 - how do we continue filling up the DONE system when we use an object
939 for the second time? "fully covered" and "uptodate" or new terminology.
941 - overview called on the wrong file should be understandable
943 - the meta data field that must change when we fake something up so that
944 the downstream people know they have to re-fetch everything.
946 - how tolerant are we against missing files upstream? how do we keep
947 track? there are legitimate cases where we did read upstream index right
948 before a file got deleted there and then find that file as new and want
949 it. There are other cases that are not self healing and must be tracked
950 and bugreported.
952 - how, exactly, do we have to deal with deletes? With rsync errors?
954 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
955 authors) failed: No such file or directory (2)
957 The file above is a delete in 1h and a new in file 1M and the
958 delete in the locally running rmirror did not get propagated to the 1M
959 object. Bug. And the consequence is a standstill.
961 It seems that a slave that works with a file below the principal needs
962 to merge things all the way up to get rid of later deletes. Or keep
963 track of all deletes and skip them later. So we need a trackdeletes.pm
964 similar to the done.pm?
966 see also 2008-08-20 about spurious deletes that really have no add
967 counterpart and yet they are not wrong.
969 - consider the effect when resyncing the recentfile takes longer than
970 the time per loop. Then we never rsync any file. We need to diagnose
971 that and force an increase of that loop time. But when we later are fast
972 enough again because the net has recovered, then we need to switch back
973 to original parameters. ERm, no, it's enough to keep syncing at least
974 one file before refetching an index file.
976 - remember to verify that no temp files are left lying around and the
977 signal handler
979 - status file for not long running jobs that want to track upstream with
980 a, say, cronjob.
982 - revisit all XXX _float areas and study Sub::Exporter DONE
984 - persistent DB even though we just said we do not need it. Just for
985 extended capabilities and time savings when, for example, upstream
986 announces a reset and we get new recentfiles and could then limit
987 ourselves to a subset of files (those that have a changed epoch) in a
988 first pass and would only then do the loop to verify the rest. Or
989 something.
991 * Todo: aggregate files should know their feed and finding the principal
992 should be done stepwise. (?)
994 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
995 left locks around. DONE
997 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
999 * maybe extend the _overview so that it always says if and where the
1000 last file is in the next file and where the next event in the next rf
1001 would lie. No, don't like this anymore. REJECT
1003 * take the two new redundant tests out again, only the third must
1004 survive. DONE
1006 * Todo: add a sanity check if the merged structure is really pointing to
1007 a different rf and that this different rf is larger. DONE
1009 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1011 * now test, if they are overlapping. And test if there is a file in the
1012 next rf that would fit into this rf's interval.
1014 1h 1222324012.8474 1222322541.7963 0.4086
1015 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
1016 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
1017 1W 1222313218.3626 1221708477.5829 167.9835
1019 I suspect that somebody writes a merged timestamp without having merged
1020 and then somebody else relies on it.
1022 If aggregate is running, the intervals must not be extravagated, if it
1023 is not running, there must not be bounds, the total number of events in
1024 the system must be counted and must be controlled throughout the tests.
1025 That the test required the additional update was probably nonsense,
1026 because aggregate can cut pieces too. FIXED & DONE
1028 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1030 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
1031 changed. FIXED
1033 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1035 * Most apparent bug at the moment is that the recentfiles are fetched
1036 too often. Only the principal should be fetched and if it has not
1037 changed, the others should not be refetched. ATM I must admit that I'm
1038 happy that we refetch more often than needed because I can more easily
1039 fix bugs while the thing is running.
1041 * Let's say, 1220474966.19501 is a timestamp of a file that is already
1042 done but the done system does not know about it. The reason for the
1043 failure is not known and we never reach the status uptodate because of
1044 this. We must get over it.
1046 Later it turns out that the origin server had a bug somewhere.
1047 1220474966.19042 came after 1220474966.19501. Or better: it was in the
1048 array of the recentfile one position above. The bug was my own.
1050 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1052 * There is the race condition where the server does a delete and the
1053 slave does not yet know and then tries to download it because he sees
1054 the new. So for this time window we must be more tolerant against
1055 failure. If we cannot download a file, we should just skip it and should
1056 not retry immediately. The whole system should discover the lost thing
1057 later. Keeping track with the DONE system should really be a no brainer.
1059 But there is something more: the whole filesystem is a database and the
1060 recentfiles are one possible representation of it. It's a pretty useful
1061 representation I think that's why I have implemented something around
1062 it. But for strictly local operation it has little value. For local
1063 operation we would much rather have a database. So we would enter every
1064 recentfile reading and every rsync operation and for every file the last
1065 state change and what it leads to. Then we would always ignore older
1066 records without the efforts involved with recentfiles.
1068 The database would have: path,recentepoch,rsyncedon,deletedon
1070 Oh well, not yet clear where this leads to.
1072 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1074 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
1075 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
1077 * statusfile, maybe only on demand, alone to have a sharp debugging
1078 tool. It is locked and all recentfiles dump themselves into it and we
1079 can build a viewer that lets us know where we stand and what's inside.
1081 * remember: only the principal recentfile needs expiration, all others
1082 shall be expired by principal if it discovers that something has move
1083 upstream.
1085 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1087 * Always check if we stringify to a higher value than in the entry
1088 before. DONE
1090 * And in covered make an additional check if we would be able to see a
1091 numerical difference between the two numbers and if we can't then switch
1092 to a different, more expensive algorithm. Do not want to be caught by
1093 floating surprises. DONE
1095 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1097 * caching has several aspects here: we can cache the interval of the
1098 recentfile which only will change when the mtime of the file changes. We
1099 must re-mirror the recentfile when its ttl has expired. Does have_read
1100 tell you anything? It counts nothing at all. Only the mtime is
1101 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
1102 thing the have_mirrored because it is unrelated to the mtime.
1104 * Robustness of floating point calculations! I always thought that the
1105 string calculated by the origin server for the floating representation
1106 of the epoch time is just a string. When we convert it to a number and
1107 later back to a string, the other computer might come to a different
1108 conclusion. This must not happen, we want to preserve it under any
1109 circumstances. I will have to write tests with overlong sequences that
1110 get lost in arithmetic and must see if all still works well. DONE
1112 But one fragile point remains: if one host considers a>b and the other
1113 one considers them == but no eq. To prevent this, we must probably do
1114 some extra homework. DONE
1116 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1118 * the concept of tracking DONE needs an object per recentfile that has
1119 something like these methods:
1121 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1123 covered() register() covered()
1125 The unclear thing is how we translate points in time into intervals. We
1126 could pass a reference to the current recent_events array when running
1127 we_have(xxx) and let the DONE object iterate over it such that it only
1128 has to store a list of intervals that can melt into each other. Ah, even
1129 passing the list together with a list of indexes seems feasiable.
1131 Or maybe ask for the inverted list?
1133 Whenever the complete array is covered by the interval we say we are
1134 fully covered and if the recentfile is not expired, we are uptodate.
1136 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1138 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1140 * need a way to "return" the next entry after the end of a list. When
1141 the caller says "before" or "after" we would like to know if he could
1142 cover that interval/threshold or not because this influences the effect
1143 of a newer timestamp of that recentfile. DONE with $opt{info}.
1145 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1147 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1149 loop implies tixu (?). Nope, something like --statefile decides. Per
1150 default we do ...?
1152 T test, I init, X including removals, U nodirtymark
1154 So we have no concept of dirtymarks, we only trust that since we are
1155 running we have observed everything steadily. But people will not let
1156 this program run forever so we must consider both startup penalty and
1157 book keeping for later runs. We keep this for later. For now we write a
1158 long running mirror that merges several intervals.
1160 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1162 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1163 the intervals!
1165 * rersyncrecent, the script: default to one week. The name of the switch
1166 is --after. Other switches? --loop!
1168 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1170 * need a switch --skip-deletes (?)
1172 * need a switch --enduser that tells us that the whole tempfile
1173 discipline is not needed when there is no downstream user. (?)
1175 Without this switch we cannot have a reasonable recent.pl that just
1176 displays the recent additions. Either we accept to download everything.
1177 Or we download temporary files without the typical rsync protocol
1178 advantages.
1180 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1181 File::Temp::tempdir, this might be a win.
1183 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1185 * apropos missing: we have no push, we never know the downstream
1186 servers. People who know their downstream hosts and want to ascertain
1187 something will want additional methods we have never thought about, like
1188 update or delete a certain file.
1190 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1192 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1193 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1194 difference.
1196 * rsync options: it's a bit of a pain that we usually need several rsync
1197 options, like compress, links, times, checksum and that there is no
1198 reasonable default except the original rsync default. I think wee can
1199 safely assume that the rsync options are shared between all recentfile
1200 instances within one recent tree.
1202 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1204 * deletes: if a delete follows an add quickly enough it may happen that
1205 a downstream mirror did not see the add at all! It seems this needs to
1206 be mentioned somewhere. The point here is that even if the downstream is
1207 never missing the principal timeframe it may encounter a "delete" that
1208 has no complimentary "add" anywhere.
1210 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1212 * I suspect the treat of metadata is incorrect during read or something.
1213 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1214 contained more than 6 hours worth of data. At 06:08 we merged into the
1215 1d file. We need to take snapshots of the 6h file over the course of an
1216 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1217 enough.
1219 Much worse: watching the 1h file: right at the moment (at 06:35) it
1220 covers 1218867584-1219120397 which is 70 hours.
1222 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1223 08:19:44 2008, that is when I checked out last time, so it seems to be
1224 aggregating and never truncating?
1226 No, correct is: it is never truncating; but wrong is: it is aggregating.
1227 It does receive a lot of events from time to time from a larger file.
1228 Somehow a large file gets merged into the small one and because the
1229 "meta/merged" attribute is missing, nobody is paying attention. I
1230 believe that I can fix this by making sure that metadata are honoured
1231 during read. DONE and test adjusted.
1233 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1235 * grand renaming plan
1237 remotebase => remoteroot to fit well with localroot DONE
1238 local_path() => localroot seems to me should already work DONE
1239 recentfile_basename => rfilename no need to stress it has no slash DONE
1241 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1242 kern? It would anyway need a deprecation cycle because it is an
1243 important constructor.
1245 * I like the portability that Data::Serializer brings us but the price
1246 is that some day we might find out that it is slowing us a bit. We'll
1247 see.
1249 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1251 * should we not enter the interval of the principal (or the interval of
1252 the merging file?) in every aggregated/merged file?
1254 * we should aim at a first release and give up on thinking about
1255 sanitizing stuff and zloop. Let's just admit that a full traditional
1256 rsync is the only available sanitizer ATM. Otherwise it's complicated
1257 stuff: sanitizing on the origin server, sanitizing on the slaves,
1258 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1259 get the basics out before this becomes a major cause for mess.
1261 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1263 * On OSes not supporting symlinks we expect that RECENT.recent contains
1264 the contents of the principal recentfile. Actually this is identical on
1265 systems supporting symlinks. Simple, what follows from that is that we
1266 need to keep the serializer in the metadata because we cannot read it
1267 from the filename, doesn't it? Of course not. It's a chicken and egg
1268 problem. This leaves us with the problem to actually parse the
1269 serialized data to find out in which format it is. So who can do the 4
1270 or 5 magics we wanted to support? File::LibMagic?
1272 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1274 * remotebase and recentfile_basename are ugly names. Now that we need a
1275 word for the shortest/principal/driving recentfile too we should do
1276 something about it.
1278 localroot is good. rfile is good. local_path() is bad, local_path($path)
1279 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1280 already deprecated.
1282 Up to now remotebase was the string that described the remote root
1283 directory in rsync notation, like pause.perl.org::authors. And
1284 recentfile_basename was "RECENT-1h.yaml".
1286 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1288 * The test that was added in today's checkin is a good start for a test
1289 of rmirror. We should have more methods in Recent.pm: verify,
1290 addmissingfiles. We should verify the current tree, then rmirror it and
1291 then verifytree the copy. We could then add some arbitrary file and let
1292 it be discovered by addmissingfiles, then rmirror again and then
1293 verifytree the copy again.
1295 Then we could start stealing from csync2 sqlite database [no port to
1296 OSX!] and fill a local DB. And methods to compare the database with the
1297 recentfiles. Our strength is that in principle we could maintain state
1298 with a single float. We have synced up to 1234567890.123456. If the Z
1299 file does not add new files all we have to do is mirror the new ones and
1300 delete the goners.
1302 This makes it clear that we should extend current protocol and declare
1303 that we cheat when we add files too late, just to help the other end
1304 keeping track. Ah yes, that's what was meant when zloop was mentioned
1305 earlier.
1307 Maybe need to revisit File::Mirror to help me with this task.
1309 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1311 * There must be an allow-me-to-truncate flag in every recentfile.
1312 Without it one could construct a sequence of updates winning the locking
1313 battle against the aggregator. Only if an aggregator has managed to
1314 merge data over to the next level, truncating can be allowed. DONE with
1315 accessor merged.
1317 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1319 * We should probably guarantee that no duplicates enter the aggregator
1320 array.
1322 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1324 * To get merge operation faster would need a good benchmark test. What
1325 02 spits out isn't reliable enough and is dominated by many other
1326 things. Between
1328 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1329 Author: Andreas J. Koenig <k@k75.(none)>
1330 Date: Sat Aug 2 07:58:04 2008 +0200
1332 and
1334 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1335 Author: Andreas J. Koenig <k@k75.(none)>
1336 Date: Sat Aug 2 11:40:29 2008 +0200
1338 there was a lot of trying but the effect is hardly measurable with
1339 current tests.
1341 * overhead of connecting seems high. When setting
1342 max_files_per_connection to 1 we see that.
1344 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1346 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1348 25138.2215600014/3600 = 6.98283932222261
1350 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1351 be a bug in the aggregator. FIXED
1353 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1355 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1356 we have it, can we mirror it, mirror it!
1358 I fear this needs a new class which might be called
1359 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1360 and treat them as an entity. I realize that a single recentfile may be
1361 sufficient for certain tasks and that it is handy for the low level
1362 programmer but it is not nice to use. If there is a delete in the 1h
1363 file then the 6h file still contains it. Seekers of the best information
1364 need to combine at least some of the recentfiles most of the time.
1366 There is the place for the Z loop!
1368 But the combination is something to collect in a database, isn't it. Did
1369 csync2 just harrumph?
1371 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1373 * it just occurred to me that hosts in the same mirroring pool could
1374 help out each other even without rewriting the recentfile. Just fetch
1375 the stuff to mirror from several places, bingo. But that's something
1376 that should rather live in a separate package or in rsync directly.
1378 * cronjobs are unsuited because with ntp they would all come at the full
1379 minute and disturb each other. Besides that I'd hate to have a backbone
1380 with more than a few seconds latency.
1382 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1384 * a second rsync server with access control for PAUSE. Port? 873 is the
1385 standard port, let's take 8873.
1387 * if there were a filesystem based on this, it would have a slow access
1388 to inexistent files. It would probably provide wrong readdir (only based
1389 on current content) or also a slow one (based on a recentfile written
1390 after the call). But it would provide fast access to existing files. Or
1391 one would deliberately allow slightly blurred answers based on some
1392 sqlite reflection of the recentfiles.
1394 * todo: write a variant of mirror() that combines two or more
1395 recentfiles and treats them like one
1397 * todo: signal handler to remove the tempfile
1399 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1401 * now that we have the symlink I forgot how it should be used in
1402 practice.
1404 * the z loop: add missing files to Z file. Just append them (instead of
1405 prepending). So one guy prepends something from the Y file from time to
1406 time and another guy appends something rather frequently. Collecting
1407 pond. When Y merges into Z, things get epoch and the collecting pond
1408 gets smaller. What exactly are "missing files"?
1410 take note of current epoch of the alpha file, let's call it the
1411 recent-ts
1413 find all files on disk
1415 remove all files registered in the recentworld up to recent-ts
1417 remove all files that have been deleted after recent-ts according to
1418 recentworld
1420 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1422 * rersyncrecent might be a cronjob with a (locked) state file which
1423 contains things like after and maybe last z sync or such?
1425 rrr-mirror might be an alternative name but how would we justify the
1426 three Rs when there is no Re-Rsync-Recent?
1428 With the --loop parameter it is an endless loop, without it is no loop.
1429 At least this is simple.
1431 * todo: new accssor z-interval specifies how often the Z file is updated
1432 against the filesystem. We probably want no epoch stamp on these
1433 entries. And we want to be able to filter the entries (e.g. no
1434 by-modules and by-category tree)
1436 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1438 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1439 into Z. We do not need the other files filled up with filesystem
1440 contents though.
1442 * need interface to query for a file in order to NOT call update on
1443 PAUSE a second time within a short time.
1445 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1447 * recommended update interval? Makes no sense, is different for
1448 different users.
1450 * Moosify
1452 Local Variables:
1453 mode: change-log
1454 change-log-default-name: "Todo"
1455 tab-width: 2
1456 left-margin: 2
1457 End: