| blob | b110dea22b8624b7b8fe4ca8f5c1171a3af4d90d |
1 2010-06-15 Andreas J. Koenig <andk@cpan.org>
3 * remove the wild alpha alerts and say, it is alpha but parts of it are
4 already working very well or so.
6 2009-12-10 Andreas J. Koenig <andk@cpan.org>
8 * Prominent request for statusfile to safe memory consumption. I'd like
9 to work backwards: get a verification program that verifies how well in
10 sync two nodes are without using frmr itself. Pretend that the forking
11 agent already works and can be interrupted with ^C. Then let it die at
12 random and restart at random and see the memory consumption and the
13 failure rate over time. Compare it with the curve we have now.
15 Of course we also invented the status file as a sharp debugging tool.
16 I'd like to see it in action.
18 2009-06-22 Andreas J. Koenig <andk@cpan.org>
20 * Todo: rrr-init to generate some index files so that rrr-server can be
21 started.
23 2009-05-02 Andreas J. Koenig <andk@cpan.org>
25 * DONE: rename rrr-dirtyupdate to rrr-update and add an option --dirty
26 so we can use it for normal update with a single file. And add a cronjob
27 that runs rrr-fsck so we get a reminder when something is not working
28 as it should.
30 2009-04-27 Andreas J. Koenig <andk@cpan.org>
32 * Todo: I still see more link_stat errors than expected. I come back to
33 the former suspicion that we do not do necessary bookkeeping on deleted
34 files so that when we reach the "new" event on a meanwhile deleted file
35 we try to mirror it although we could know better. It's just noise, no
36 harm but noise is irritating.
38 * Todo: more experimenting with runstatusfile to get it official
40 * Todo: make sure $ENV{LANG}="C" when we call rsync because we will
41 parse the error.
43 rsync: link_stat "/authors/id/T/TH/THINC/DateTime-Format-Flexible-0.02.tar.gz" (in PAUSE) failed: No such file or directory
45 rsync: link_stat "/authors/id/A/AG/AGENT/SSH-Batch-0.001.meta" (in PAUSE) failed: No such file or directory (2)
47 And that ignore_link_stat_error is propagated in sparse_clone.
49 DONE
51 For the record: the bug was that ignore_link_stat_error was lost during
52 sparse_clone. And then the second bug was that it did not default to
53 true. Both are now fixed.
55 2009-04-25 Andreas J. Koenig <andk@cpan.org>
57 * Todo: fill rrr program with life
59 2009-04-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
61 * Minor bug: I see complaints about files not existing like
62 /authors/id/G/GW/GWADEJ/SVG-Sparkline-0.2.0.meta or
63 /authors/id/R/RC/RCAPUTO/POE-1.004.readme
65 In the RECENT files these are delete events but of course also new
66 events in older recent files.
68 I think we do not properly remember deletes when we skip-deletes and so
69 do not filter them out when we later see the "new" event.
71 Yes, but that what ignore_link_stat_error is for, we know about this
72 race condition. We must make ignore_link_stat_error default to true.
74 DONE
76 * bug in Done.pm failing to merge two fields into one when a third is
77 present. FIXED and accompanied with new test.
79 2009-04-24 Andreas J. Koenig <andk@cpan.org>
81 * Todo: keep the index clean and run some kind of nondestructive fsck 4
82 times a day.
84 * who is our backbone?
86 http://cpan.cpantesters.org/authors/02STAMP # barbie
87 http://cpan.solfo.com/authors/02STAMP # abh
88 (http|ftp)://theoryx5.uwinnipeg.ca/pub/CPAN/ # rkobes
90 2009-04-17 Andreas J. Koenig <andk@cpan.org>
92 * the changelog helper I included in the Makefile release memo does not
93 show tags and is much less useful than I thought. gitk is probably much
94 more convenient.
96 2009-04-16 Andreas J. Koenig <andk@cpan.org>
98 * ABH writes:
100 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
101 #### Ydr_.yaml ... DONE
102 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
103 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
104 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
105 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
106 #### gone already at cpan-pause.pl line 0
108 Running without skip-deletes now but cannot reproduce.
110 2009-04-15 Andreas J. Koenig <andk@cpan.org>
112 * consider state directory such we can restart after a ^C or come again
113 from a cronjob.
115 Reuse _runstatusfile: write atomically, see how we can use the dump to
116 start running again. Do not leak and remove globs before writing. Do not
117 only dump one rf, dump the whole complex.
119 * consider the effect of ^C. Is it safe?
121 * want to use Perl::Version to up the versions on every release. Only if
122 things have changed, of course. Does this actually work?
124 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
125 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
127 Must do the bookkeeping myself when this should happen, maybe based on git?
129 I think I prefer the same version number for all pm files. But one also
130 needs a routine for setting them. Ahh:
132 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
134 NEEDSMOREWORK
136 * why indexer wrong?
138 cpan[2]> m /Mirror::Recent/
139 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
140 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
141 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
142 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
143 4 items found
145 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
147 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
149 I just tried MM 6.5102 and it again produces only a provides for Recent,
150 not for the other modules.
152 My own fault. FIXED in the Makefile.PL.
154 2009-04-13 Andreas J. Koenig <andk@cpan.org>
156 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
158 * is it true (as stated at
159 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
160 that rsync can lead to errors when upstream changes before client sync
161 has completed? Can we see the error when we run this on upstream:
163 % perl -e '
164 use Time::HiRes qw(time);
165 while (){
166 open my $fh, ">", "changingfile.txt.new" or die;
167 print $fh (time."\n") x 1000000;
168 close $fh;
169 rename "changingfile.txt.new", "changingfile.txt" or die;
170 }
171 '
173 And on the receiving end:
175 % while true; do
176 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
177 sleep 1;
178 done
180 I cannot get it to fail. Apparently it is sufficient when upstream
181 always writes atomically (which of course is mandatory)?
183 SENT to mailing list Apr 28.
185 * from ABH:
187 > https://fedorahosted.org/InstantMirror/
188 > https://www.redhat.com/mailman/listinfo/instantmirror-list
189 > irc.freenode.net channel #instantmirror
191 A cool name for a project. Inspires me to write a few sentences about
192 bittorrent's role in the grand picture.
194 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
195 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
197 2009-04-12 Andreas J. Koenig <andk@cpan.org>
199 * Interesting last minute bug during real download testing: the output
200 isn't as pretty anymore, it seems that more work is being done than
201 needed.
203 Introducing a highlevel _alluptodate method helped a lot in debugging
204 this.
206 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
207 different on 1h file than on the other files. This lead to frequent
208 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
209 issue.
211 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
213 * Todo: try if we can be faster with a native float package. I'm really
214 glad to have a machine exposing perl floating point bugs, it forced me
215 to retract this item from the changes file: "several speedups in the
216 fakebigfloat code" before I get cpan testers fails.
218 2009-04-11 Andreas J. Koenig <andk@cpan.org>
220 * Bug: when the dirtymark changes upstream, then the downstream server
221 notices it immediately and mirrors the "1h" file but then leaves the
222 rmirror loop and continues with "6h" on the next round through. This
223 behaviour goes on until it reaches the Z file. Because we eat the Z file
224 piecemeal we leave the loop after a while but after that everybody seems
225 to have forgotten that there is some work left to be done. Bad, bad,
226 bad.
228 ETOOCONFUSING better name?
230 Recentfile::get_remote_recentfile_as_tempfile OK
231 Recentfile::resolve_recentfilename split_rfilename DONE
233 Recent::_principal_recentfile_object _principal_recentfile_fromremote
234 Recent::principal_recentfile OK
235 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
237 some void, some not, some not void but called in void context.
239 2009-04-10 Andreas J. Koenig <andk@cpan.org>
241 * Bug? When the dirtymark changes then the second tier hosts quickly
242 reset their DONE state and restart mirroring. But then they mirror
243 potentially outdated index files with inconsistent dirtymark because the
244 first tier box may be a bit behind on the large recentfiles.
246 This means we need a brake when iterating over the recentfiles that
247 refuses to use a recentfile with the wrong dirtymark.
249 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
251 * install the bin/* files and talk about rrr-overview somewhere.
253 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
254 created outside the target tree.
256 * Barbie asks for better logging capabilities
258 2009-03-29 Andreas J. Koenig <andk@cpan.org>
260 * known bug: cannot configure the Recentfile objects to keep delete
261 events. How to call the accessor for that? keep_delete_objects_forever?
263 Sounds acceptable. FIXED (but untested).
265 * known bug: keeps temporary index files lying around.
267 2009-03-24 Andreas J. Koenig <andk@cpan.org>
269 * Some equivalent for
271 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
273 ? DONE
275 2009-03-22 Andreas J. Koenig <andk@cpan.org>
277 * k81 is again client of k75 with different parameters.
279 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
280 about it all the time and nobody reacts accordingly; (2) lots of temp
281 files get created and not removed; culprit the new call to
282 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
283 caller has to remove the tempfile after use.
285 Need the drawing board.
287 PARTLY FIXED: retracting the idea to call
288 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
289 and unseed such that uptodate now means "we have mirrored this
290 recentfile's bunch of files" and seeded means "somebody believes the
291 index file for this rf needs be refreshed". This seems to work now.
293 Still collecting temp files that nobody cares to depollute.
295 2009-03-21 Andreas J. Koenig <andk@cpan.org>
297 * Bug?: should it be harder than it is atm to set the timestamp to the
298 future? YES, FIXED
300 * bug with native integers:
302 -
303 epoch: 997872011
304 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
305 type: new
306 -
307 epoch: 1195248431
308 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
309 type: new
311 Native integer broke when native math was turned off. FIXED
313 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
314 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
315 in this area.
317 Yes, records out of order:
319 447003 -
320 447004 epoch: 995885533
321 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
322 447006 type: new
323 447007 -
324 447008 epoch: 995890358
325 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
326 447010 type: new
327 447011 -
328 447012 epoch: 995892221
329 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
330 447014 type: new
332 FIXED with sanity check and later with the integer fix.
334 * Bug: want the index files in a .recent directory
336 * Bug: lots of dot files are not deleted in time
338 * possible test case: can a delete change the timestamp? This would
339 probably break the order of events.
341 2009-03-20 Andreas J. Koenig <andk@cpan.org>
343 * 1233701831.34486 what's so special about this number/string? It
344 belongs to
345 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
346 atm lives in Y,Q, and Z.
348 It is the first entry after id/--skip-locking which has timestamp
349 1234164228.11325 which represents a file that doesn't exist anymore.
351 * another bug:
353 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
354 _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
355 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
356 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
357 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
358 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
359 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
360 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
361 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
362 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
363 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
364 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
365 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
367 FIXED, it was the "--skip-locking" file where manual intervention was participating
369 * bug on the mirroring slave: when the dirtymark gets increased we
370 probably do not reset the done intervals. The mirrorer stays within
371 tight bounds where it tries to sync with upstream and never seems to
372 finish. In the debugging state file I see lots of identical intervals
373 that do not get collapsed. When I restart the mirrorer it dies with:
375 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
376 _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
377 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
378 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
379 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
380 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
381 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
382 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
383 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
384 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
386 and debugging stands at
388 At: before
389 Brfinterval: Z
390 Ci: 227
391 Dre+1:
392 epoch: 1237400802.5789
393 path: id/J/JO/JOHND/CHECKSUMS
394 type: new
395 Dre-0:
396 epoch: 1237400807.97514
397 path: id/M/MI/MIYAGAWA/CHECKSUMS
398 type: new
399 Dre-1:
400 epoch: 1237400817.94363
401 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
402 type: new
403 Eintervals:
404 -
405 - 900644040
406 - 900644040
407 - []
409 and it is reproducable.
411 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
412 pause has a fresh one.
414 FIXED, it was the third anded term in each of the ifs in the IV block in
415 _register_one: with that we make sure that we do not stamp on valuable
416 interval data.
418 2009-03-17 Andreas J. Koenig <andk@cpan.org>
420 * done: verified the existence of the floating point bug in bleadperl
421 and verified that switching from YAML::Syck to YAML::XS does not resolve
422 it.
424 BTW, the switch was doable with
426 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
428 and should be considered as a separate TODO
430 * todo: integrate a dirty update with two aggregate calls before
431 unlocking for frictionless dirtying DONE
433 * todo: start the second rsync daemon on pause DONE
435 * todo: move index files to .recent: this cannot simply be done by
436 setting filenameroot to .recent/RECENT. Other parts of the modules rely
437 on the fact that dirname(recentfile) is the root of the mirrored tree.
439 2009-03-16 Andreas J. Koenig <andk@cpan.org>
441 * What was the resolution of the mirror.pl delete hook bug? Do we call
442 the delete hook when pause removes a file from MUIR?
444 * Today on pause: Updating 2a13fba..29f284d and installing it for
445 /usr/local/perl-5.10.0{,-RC2}
447 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
448 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
450 2009-03-15 Andreas J. Koenig <andk@cpan.org>
452 * currently recent_events has the side effect of setting dirtymark
453 because it forces a read on the file. That should be transparent, so
454 that the dirtymark call always forces a cache-able(?) read.
456 * The bug below is -- after a lot of trying -- not reproducible on a
457 small script, only in the large test script. The closest to the output
458 below was:
460 #!perl
461 use strict;
462 use Devel::Peek;
463 my $x = "01237123229.8814";
464 my($l,$r);
465 for ($l,$r) {
466 $_ = "x"x34;
467 }
468 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
469 $r = int $r;
470 $l = "1237123231.22458";
471 $r = "1237123231.22458";
472 1 if $l/1.1;
473 Devel::Peek::Dump $l;
474 Devel::Peek::Dump $r;
475 Devel::Peek::Dump $x = $l <=> $r;
476 die "BROKE" if $x;
477 __END__
479 The checked in state at c404a85 fails the test with my
480 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
481 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
482 in the test script always consistent.
484 * Todo: write a test that inserts a second dirty file with an already
485 existing timestamp. DONE
487 * Bug in perl 5.10 on my 64bit box:
489 DB<98> Devel::Peek::Dump $l
490 SV = PVMG(0x19e0450) at 0x142a550
491 REFCNT = 2
492 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
493 IV = 0
494 NV = 1237123231.22458
495 PV = 0x194ce70 "1237123231.22458"\0
496 CUR = 16
497 LEN = 40
499 DB<99> Devel::Peek::Dump $r
500 SV = PVMG(0x19e0240) at 0x142a3e8
501 REFCNT = 2
502 FLAGS = (PADMY,POK,pPOK)
503 IV = 1237123229
504 NV = 1237123229.8814
505 PV = 0x19ff900 "1237123231.22458"\0
506 CUR = 16
507 LEN = 40
509 DB<100> Devel::Peek::Dump $l <=> $r
510 SV = IV(0x19ea6e8) at 0x19ea6f0
511 REFCNT = 1
512 FLAGS = (PADTMP,IOK,pIOK)
513 IV = -1
515 DB<101> Devel::Peek::Dump $l
516 SV = PVMG(0x19e0450) at 0x142a550
517 REFCNT = 2
518 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
519 IV = 0
520 NV = 1237123231.22458
521 PV = 0x194ce70 "1237123231.22458"\0
522 CUR = 16
523 LEN = 40
525 DB<102> Devel::Peek::Dump $r
526 SV = PVMG(0x19e0240) at 0x142a3e8
527 REFCNT = 2
528 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
529 IV = 1237123231
530 NV = 1237123231.22458
531 PV = 0x19ff900 "1237123231.22458"\0
532 CUR = 16
533 LEN = 40
535 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
537 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
539 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
541 * $obj->merge ($other) needs to learn about equal epoch which may happen
542 since dirty_epoch intruded.
544 * Wontfix anytime soon: I think we currently do not support mkdir. Only
545 files!
547 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
549 * Todo: continue working on update(...,$dirty_epoch). It must be
550 followed by a fast_aggregate! DONE
552 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
554 * maybe we need a closest_entry or fitting_interval or something like
555 that. We want to merge an event into the middle of some recentfile.
556 First we do not know which file, then we do not know where to lock,
557 where to enter the new item, when and where to correct the dirtymark.
559 So my thought is we should first find which file.
561 Another part of my brain answers: what would happen if we would enter
562 the new file into the smallest file just like an ordinary new event,
563 just as an old event?
565 (1) we would write a duplicate timestamp? No, this would be easy to
566 avoid
568 (2) we would make the file large quickly? Yes, but so what? We are
569 changing the dirtymark, so are willing to disturb the downstream hosts.
571 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
573 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
575 leak!
577 https://rt.cpan.org/Ticket/Display.html?id=41199
579 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
581 Strangest output being HAKANARDO who managed to upload
583 Here is a better oneliner that includes also the first line of each
584 finding:
586 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
588 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
590 * memory leak in the syncher? It currently weighs 100M.
592 Update 2008-11-02:
594 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
597 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
599 * lookup by epoch and by path and use this ability on the pause to never
600 again register a file twice that doesn't need it. Let's call it
601 contains().
603 * after the dirtymark is done: fill up recentfiles with fake (historic)
604 entries; fill up with individual corrections; algorithm maybe to be done
605 with bigfloat so that we can always place something in the middle
606 between two entries. Before we must switch to bigfloat we could try to
607 use Data::Float::nextup to get the.
609 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
610 directly.
612 * dirtymark now lives in Recentfile, needs to be used in rmirror.
614 * find out why the downloader died after a couple of hours without a net
615 connection. Write a test that survives the not-existence of the other
616 end forever.
618 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
620 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
621 it makes less sense with tight coupling for secondary files. But asking
622 the server what the current epoch is might be cheaper on HTTP than on
623 rsync. (Needs to be evaluated)
625 * remove the 0.00 from the verbose overview in the Merged column in the
626 Z row. DONE
628 * write tests that expose the problems of the last few days: cascading
629 client/server roles, tight coupling for secondary RFs, deletes after
630 copies.
632 * Some day we might want to have policy options for the slave:
633 tight/loose/no coupling with upstream for secondary RFs. tight is what
634 we have now. loose would wait until a gap occurs that can be closed.
636 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
638 * revisit all $rfs->[$i+1] places if they now make sense still
640 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
642 * another bug is the fact that the mirror command deletes files before
643 it unhides the index file, thus confusing downstream slaves. We must not
644 delete before unhiding and must delete after unhiding. FIXED.
646 * new complication about the slave that is playing a server role.
647 Currently we mirror from newest to oldest with a hidden temporary file
648 as index. And when one file is finished, we unhide the index file.
649 Imagine the cascading server/slave is dead for a day. It then starts
650 mirroring again with the freshest thing and unhides the freshest index
651 file when it has worked through it. In that moment it exposes a time
652 hole. Because it now works on the second recentfile which is still
653 hidden.
655 We currently do nothing special to converge after such a drop out. At
656 least not intentionally and robustly and thought through.
658 The algorithm we use to seed the next file needs quite a lot of more
659 robustness than it currently has. Something to do with looking at the
660 merged element of the next rf and when it has dropped off, we seed
661 immediately. And if it ramains dropped off, we seed again, of course.
663 Nope, looking from smaller to larger RFS we look at the merged element
664 of this RF and at the minmax/max element of the next RF. If that
665 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
666 seeding it.
668 And we need a public accessor seed and unseed or seeded. But not the mix
669 of public and private stuff that then is used behind the back.
671 And then the secondary* stuff must go.
673 And we must understand what the impact is on the DONE system. Can it go
674 unnoticed that there was a hole? And could the DONE system have decided
675 the hole is covered? This should be testable with three directories where
676 the middle stops working for a while. Done->merge is suspicious, we must
677 stop it from merging non-conflatable neighbors due to broken continuity.
679 FIXED
681 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
683 * Slaven suggests to have the current epoch or the whole current
684 recentfile available from the HTTP server and take it away with
685 keepalive. This direction goes the granularity down to subseconds.
687 We might want to rewrite everything to factor out transport and allow
688 the whole thing to run via HTTP.
690 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
692 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
693 symlink?
695 * smoker on k81 fetching from k75 to verify cascading works. See
696 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
698 * maybe the loop should wait for CHECKSUMS file after every upload. And
699 CPAN.pm needs to deal with timestamps in the future.
701 * do not forget the dirtymark!
703 Text: have a new flag on recentfiles with the meaning: if this
704 changes, you're required to run a full rsync over all the files. The
705 reason why we set it would probably be: some foul happened. we injected
706 files in arbitrary places or didn't inject them although they changed.
707 The content of the flag? Timestamp? The relation between the
708 recentfiles would have to be inheritance from the principal, because any
709 out of band changes would soon later propagate to the next recentfile.
711 By upping the flag often one can easily ruin the slaves.
713 last out of band change? dirtymark?
715 Anyway, this implies that we read a potentially existing recentfile
716 before we write one.
718 And it implies that we have an eventloop that keeps us busy in 2-3
719 cycles, one for current stuff (tight loop) and one for the recentfiles
720 (cascade when principal has changed), one for the old stuff after a
721 dirtymark change.
723 And it implies that the out-of-band change in any of the recentfiles
724 must have a lock on the principal file and there is the place to set the
725 dirtymark.
727 * start a FAQ, especially quick start guide questions. Also to aid those
728 problematic areas where we have no good solution, like the "links"
729 option to rsync.
731 * wish feedback when we are slow.
733 * reduce mccabe
735 * Remove a few DEBUG statements.
737 * The multiple-rrr way of doing things needs a new option to rmirror,
738 like piecemeal or so. Not urgent because after the first pass through,
739 things run smoothely. It's only ugly during the first pass.
741 * I have the suspicion that the code is broken that decides if the
742 neighboring RF needs to be seeded. I fear when too much time has gone
743 between two calls (in our case more than one hour), it would not seed
744 the neighbor. Of course this will never be noticed, so we need a good
745 test for it.
747 * local/localroot confusion: I currently pass both options but one must
748 do.
750 * accounts for early birds on PAUSE rsync daemon.
752 * hardcoded 20 seconds
754 * who mirrors the index? DOING now.
756 * which CPAN mirrors offer rsync?
758 * visit all XXX, visit all _float places
760 * rename the pathdb stuff, it's too confusing. No idea how.
762 * rrr-inotify, backpan, rrr-register
764 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
766 * current bugs: the pathdb seems to get no reset, the seeding of the
767 secondaryttl stuff seems not to have an effect. Have helped myself with
768 a rand(10), need to fix this back. So not checked in. Does the rand
769 thing even help?
771 The rand thing helps. The secondaryttl stuff was in the wrong line,
772 fixed now.
774 The pathdb stuff was because I called either _pathdb or __pathdb on the
775 wrong object. FIXED now.
777 * It's not so beautiful if we never fetch the recentfiles that are not
778 the principal, even if this is correct behaviour. We really do not need
779 them after we have fetched the whole content.
781 OK, we want a switch for that: secondaryttl DONE
783 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
785 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
786 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
787 it lets 40 follow. FIXED
789 * See that the long running process really only updates the principal
790 file unless it has missed a timespan during which something happened. If
791 nothing happened, it must notice even when it misses the timespan. DONE
793 * we must throw away the pathdb when we have reached the end of Z. From
794 that moment we can have a very small pathdb because the only reason for
795 a pathdb is that we know to ignore old records in old files. We won't
796 need this pathdb again before the next full pass over the data is
797 necessary and then we will rebuild it as we go along. DONE
799 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
801 * I think, Done::register_one is doing wrong in that it does not
802 conflate neighboring pieces. The covered() method cannot do this because
803 it has no recent_events array at hand. But register_one has it and could
804 do it and for some reason misses to do it (sometimes).
806 This means that the three tests I just wrote can probably not survive
807 because they test with an already broken Done structure.
809 The art now is to detect how it happens, then to reproduce, then write a
810 test, then fix it.
812 So from the logfile this is what happens: we have a good interval with
813 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
814 top of it. Locally we now mirror F2 and open a new done interval for it.
815 Then we mirror F1 but this time with the timestamp T1b. And when we then
816 try to close the gap, we do not find T1 but instead something older. We
817 should gladly accept this older piece and this would fix this bug.
819 FIXED
821 * bug to fix: when the 1h file changes while rmirror is running, we do
822 correctly sync the new files but never switch to the 6h file but rather
823 stay in a rather quick loop that fetches the 1h file again and again.
825 Is it possible that we initialize a new object? Or does
826 get_remote_recentfile_as_tempfile overwrite something in myself?
828 Want a new option: _runstatusfile => $file which frequently dumps the
829 state of all recentfiles to a file.
831 FIXED
833 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
835 * Todo: now teach update to verify the timestamp is about to write
836 against the previous and use _increase_a_bit if it doesn't comply with
837 strict monotony. DONE
839 * The problem of rounding. So far perl's default precision was
840 sufficient. One day it won't be. FakeFloat has an easy job when it is
841 only reading and other machines have written correctly. But when we want
842 to write a floating point number that is a bit larger than the other
843 one, then we need our own idea of precision.
845 Slaven said: just append a "1". This might be going towards the end of
846 usability too quickly. I'd like something that actually uses the decimal
847 system. Well, appending a 1 also does this but...
849 E.g. we have 1.0. nextup on this architecture is starting with
850 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
851 taken the 1.0000000000000003 and the next user comes and the time tells
852 him 1.0 again. He has to beat my number without stepping over the
853 nextup. This is much less space than I had when I chose 1,2,3.
855 What is also irritating is that nextup is architecture dependent. The
856 128 bit guy must choose very long numbers to fit in between whereas the
857 other one with 16 bit uses larger steps. But then the algorithm is the
858 same for both, so that would be a nice thing.
860 I see two situation where we need this. One is when Time::HiRes returns
861 us a value that is <= the last entry in our recentfile. In this case
862 (let's call it the end-case) we must fill the region between that number
863 and the next higher native floating point number. The other is when we
864 inject an old file into an old recentfile (we would then also set a new
865 dirtymark). We find the integer value already taken and need a slightly
866 different one (let's call it the middle-case). The difference between
867 the two situations is that the next user will want to find something
868 higher than my number in the end-case and something lower than my number
869 in the middle case.
871 So I suggest we give the function both a value and an upper bound and it
872 calculates us a primitive middle. The upper bound in the middle-case is
873 the next integer. The upper bound on the end-case is the nextup floating
874 point number. But the latter poses another problem: if we have occupied
875 the middle m between x and nextup(x), then the nextup(m) will probably
876 not be the same as nextup(x) because some rounding will take place
877 before the nextup is calculated and when the rounding reaches the
878 nextup(x), we will end up at nextup(nextup(x)).
880 So we really need to consider the nextup and the nextdown from there and
881 then the middle and that's the number we may approach asymptotically.
882 Ugly. But DONE.
884 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
886 * consider deprecating the use of RECENT.recent as a symlink. It turns
887 out to need extra hoops with the rsync options and just isn't worth it.
888 Or maybe these extra hoops are needed anyway for the rest of the tree?
889 Nope, can't be the case because not all filesystems support symlinks.
891 But before doing the large step, I'll deprecate the call of
892 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
893 only introduced to resolve RECENT.recent and complicates the routine far
894 beyond what it deserves.
896 DONE. Won't deprecate RECENT.recent, just moved its handling to the
897 supervisor.
899 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
901 * I think it's a bug that the rsync_option links must be set to true in
902 order to support RECENT.recent and that nobody cares to set it
903 automatically. Similar for ignore_link_stat_errors.
905 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
907 * Todo: find all todos together and make a plan what is missing for a
908 release.
910 - verifytree or something like that. fsck maybe.
912 - rersyncrecent, the script itself? What it do?
914 - a way to only mirror the recentfiles without mirroring the whole
915 remote system such that people can decide to mirror only partially see
916 also 2008-08-30. .shadow-xxx directory? this also needed for a
917 filesystem that is still incomplete and might need the mirrorfiles for
918 lookup(?)
920 - long living objects that mirror again and again. Inject something
921 into ta, see how it goes over to tb.
923 - how do we continue filling up the DONE system when we use an object
924 for the second time? "fully covered" and "uptodate" or new terminology.
926 - overview called on the wrong file should be understandable
928 - the meta data field that must change when we fake something up so that
929 the downstream people know they have to re-fetch everything.
931 - how tolerant are we against missing files upstream? how do we keep
932 track? there are legitimate cases where we did read upstream index right
933 before a file got deleted there and then find that file as new and want
934 it. There are other cases that are not self healing and must be tracked
935 and bugreported.
937 - how, exactly, do we have to deal with deletes? With rsync errors?
939 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
940 authors) failed: No such file or directory (2)
942 The file above is a delete in 1h and a new in file 1M and the
943 delete in the locally running rmirror did not get propagated to the 1M
944 object. Bug. And the consequence is a standstill.
946 It seems that a slave that works with a file below the principal needs
947 to merge things all the way up to get rid of later deletes. Or keep
948 track of all deletes and skip them later. So we need a trackdeletes.pm
949 similar to the done.pm?
951 see also 2008-08-20 about spurious deletes that really have no add
952 counterpart and yet they are not wrong.
954 - consider the effect when resyncing the recentfile takes longer than
955 the time per loop. Then we never rsync any file. We need to diagnose
956 that and force an increase of that loop time. But when we later are fast
957 enough again because the net has recovered, then we need to switch back
958 to original parameters. ERm, no, it's enough to keep syncing at least
959 one file before refetching an index file.
961 - remember to verify that no temp files are left lying around and the
962 signal handler
964 - status file for not long running jobs that want to track upstream with
965 a, say, cronjob.
967 - revisit all XXX _float areas and study Sub::Exporter DONE
969 - persistent DB even though we just said we do not need it. Just for
970 extended capabilities and time savings when, for example, upstream
971 announces a reset and we get new recentfiles and could then limit
972 ourselves to a subset of files (those that have a changed epoch) in a
973 first pass and would only then do the loop to verify the rest. Or
974 something.
976 * Todo: aggregate files should know their feed and finding the principal
977 should be done stepwise. (?)
979 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
980 left locks around. DONE
982 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
984 * maybe extend the _overview so that it always says if and where the
985 last file is in the next file and where the next event in the next rf
986 would lie. No, don't like this anymore. REJECT
988 * take the two new redundant tests out again, only the third must
989 survive. DONE
991 * Todo: add a sanity check if the merged structure is really pointing to
992 a different rf and that this different rf is larger. DONE
994 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
996 * now test, if they are overlapping. And test if there is a file in the
997 next rf that would fit into this rf's interval.
999 1h 1222324012.8474 1222322541.7963 0.4086
1000 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
1001 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
1002 1W 1222313218.3626 1221708477.5829 167.9835
1004 I suspect that somebody writes a merged timestamp without having merged
1005 and then somebody else relies on it.
1007 If aggregate is running, the intervals must not be extravagated, if it
1008 is not running, there must not be bounds, the total number of events in
1009 the system must be counted and must be controlled throughout the tests.
1010 That the test required the additional update was probably nonsense,
1011 because aggregate can cut pieces too. FIXED & DONE
1013 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1015 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
1016 changed. FIXED
1018 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1020 * Most apparent bug at the moment is that the recentfiles are fetched
1021 too often. Only the principal should be fetched and if it has not
1022 changed, the others should not be refetched. ATM I must admit that I'm
1023 happy that we refetch more often than needed because I can more easily
1024 fix bugs while the thing is running.
1026 * Let's say, 1220474966.19501 is a timestamp of a file that is already
1027 done but the done system does not know about it. The reason for the
1028 failure is not known and we never reach the status uptodate because of
1029 this. We must get over it.
1031 Later it turns out that the origin server had a bug somewhere.
1032 1220474966.19042 came after 1220474966.19501. Or better: it was in the
1033 array of the recentfile one position above. The bug was my own.
1035 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1037 * There is the race condition where the server does a delete and the
1038 slave does not yet know and then tries to download it because he sees
1039 the new. So for this time window we must be more tolerant against
1040 failure. If we cannot download a file, we should just skip it and should
1041 not retry immediately. The whole system should discover the lost thing
1042 later. Keeping track with the DONE system should really be a no brainer.
1044 But there is something more: the whole filesystem is a database and the
1045 recentfiles are one possible representation of it. It's a pretty useful
1046 representation I think that's why I have implemented something around
1047 it. But for strictly local operation it has little value. For local
1048 operation we would much rather have a database. So we would enter every
1049 recentfile reading and every rsync operation and for every file the last
1050 state change and what it leads to. Then we would always ignore older
1051 records without the efforts involved with recentfiles.
1053 The database would have: path,recentepoch,rsyncedon,deletedon
1055 Oh well, not yet clear where this leads to.
1057 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1059 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
1060 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
1062 * statusfile, maybe only on demand, alone to have a sharp debugging
1063 tool. It is locked and all recentfiles dump themselves into it and we
1064 can build a viewer that lets us know where we stand and what's inside.
1066 * remember: only the principal recentfile needs expiration, all others
1067 shall be expired by principal if it discovers that something has move
1068 upstream.
1070 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1072 * Always check if we stringify to a higher value than in the entry
1073 before. DONE
1075 * And in covered make an additional check if we would be able to see a
1076 numerical difference between the two numbers and if we can't then switch
1077 to a different, more expensive algorithm. Do not want to be caught by
1078 floating surprises. DONE
1080 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1082 * caching has several aspects here: we can cache the interval of the
1083 recentfile which only will change when the mtime of the file changes. We
1084 must re-mirror the recentfile when its ttl has expired. Does have_read
1085 tell you anything? It counts nothing at all. Only the mtime is
1086 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
1087 thing the have_mirrored because it is unrelated to the mtime.
1089 * Robustness of floating point calculations! I always thought that the
1090 string calculated by the origin server for the floating representation
1091 of the epoch time is just a string. When we convert it to a number and
1092 later back to a string, the other computer might come to a different
1093 conclusion. This must not happen, we want to preserve it under any
1094 circumstances. I will have to write tests with overlong sequences that
1095 get lost in arithmetic and must see if all still works well. DONE
1097 But one fragile point remains: if one host considers a>b and the other
1098 one considers them == but no eq. To prevent this, we must probably do
1099 some extra homework. DONE
1101 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1103 * the concept of tracking DONE needs an object per recentfile that has
1104 something like these methods:
1106 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1108 covered() register() covered()
1110 The unclear thing is how we translate points in time into intervals. We
1111 could pass a reference to the current recent_events array when running
1112 we_have(xxx) and let the DONE object iterate over it such that it only
1113 has to store a list of intervals that can melt into each other. Ah, even
1114 passing the list together with a list of indexes seems feasiable.
1116 Or maybe ask for the inverted list?
1118 Whenever the complete array is covered by the interval we say we are
1119 fully covered and if the recentfile is not expired, we are uptodate.
1121 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1123 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1125 * need a way to "return" the next entry after the end of a list. When
1126 the caller says "before" or "after" we would like to know if he could
1127 cover that interval/threshold or not because this influences the effect
1128 of a newer timestamp of that recentfile. DONE with $opt{info}.
1130 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1132 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1134 loop implies tixu (?). Nope, something like --statefile decides. Per
1135 default we do ...?
1137 T test, I init, X including removals, U nodirtymark
1139 So we have no concept of dirtymarks, we only trust that since we are
1140 running we have observed everything steadily. But people will not let
1141 this program run forever so we must consider both startup penalty and
1142 book keeping for later runs. We keep this for later. For now we write a
1143 long running mirror that merges several intervals.
1145 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1147 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1148 the intervals!
1150 * rersyncrecent, the script: default to one week. The name of the switch
1151 is --after. Other switches? --loop!
1153 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1155 * need a switch --skip-deletes (?)
1157 * need a switch --enduser that tells us that the whole tempfile
1158 discipline is not needed when there is no downstream user. (?)
1160 Without this switch we cannot have a reasonable recent.pl that just
1161 displays the recent additions. Either we accept to download everything.
1162 Or we download temporary files without the typical rsync protocol
1163 advantages.
1165 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1166 File::Temp::tempdir, this might be a win.
1168 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1170 * apropos missing: we have no push, we never know the downstream
1171 servers. People who know their downstream hosts and want to ascertain
1172 something will want additional methods we have never thought about, like
1173 update or delete a certain file.
1175 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1177 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1178 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1179 difference.
1181 * rsync options: it's a bit of a pain that we usually need several rsync
1182 options, like compress, links, times, checksum and that there is no
1183 reasonable default except the original rsync default. I think wee can
1184 safely assume that the rsync options are shared between all recentfile
1185 instances within one recent tree.
1187 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1189 * deletes: if a delete follows an add quickly enough it may happen that
1190 a downstream mirror did not see the add at all! It seems this needs to
1191 be mentioned somewhere. The point here is that even if the downstream is
1192 never missing the principal timeframe it may encounter a "delete" that
1193 has no complimentary "add" anywhere.
1195 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1197 * I suspect the treat of metadata is incorrect during read or something.
1198 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1199 contained more than 6 hours worth of data. At 06:08 we merged into the
1200 1d file. We need to take snapshots of the 6h file over the course of an
1201 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1202 enough.
1204 Much worse: watching the 1h file: right at the moment (at 06:35) it
1205 covers 1218867584-1219120397 which is 70 hours.
1207 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1208 08:19:44 2008, that is when I checked out last time, so it seems to be
1209 aggregating and never truncating?
1211 No, correct is: it is never truncating; but wrong is: it is aggregating.
1212 It does receive a lot of events from time to time from a larger file.
1213 Somehow a large file gets merged into the small one and because the
1214 "meta/merged" attribute is missing, nobody is paying attention. I
1215 believe that I can fix this by making sure that metadata are honoured
1216 during read. DONE and test adjusted.
1218 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1220 * grand renaming plan
1222 remotebase => remoteroot to fit well with localroot DONE
1223 local_path() => localroot seems to me should already work DONE
1224 recentfile_basename => rfilename no need to stress it has no slash DONE
1226 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1227 kern? It would anyway need a deprecation cycle because it is an
1228 important constructor.
1230 * I like the portability that Data::Serializer brings us but the price
1231 is that some day we might find out that it is slowing us a bit. We'll
1232 see.
1234 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1236 * should we not enter the interval of the principal (or the interval of
1237 the merging file?) in every aggregated/merged file?
1239 * we should aim at a first release and give up on thinking about
1240 sanitizing stuff and zloop. Let's just admit that a full traditional
1241 rsync is the only available sanitizer ATM. Otherwise it's complicated
1242 stuff: sanitizing on the origin server, sanitizing on the slaves,
1243 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1244 get the basics out before this becomes a major cause for mess.
1246 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1248 * On OSes not supporting symlinks we expect that RECENT.recent contains
1249 the contents of the principal recentfile. Actually this is identical on
1250 systems supporting symlinks. Simple, what follows from that is that we
1251 need to keep the serializer in the metadata because we cannot read it
1252 from the filename, doesn't it? Of course not. It's a chicken and egg
1253 problem. This leaves us with the problem to actually parse the
1254 serialized data to find out in which format it is. So who can do the 4
1255 or 5 magics we wanted to support? File::LibMagic?
1257 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1259 * remotebase and recentfile_basename are ugly names. Now that we need a
1260 word for the shortest/principal/driving recentfile too we should do
1261 something about it.
1263 localroot is good. rfile is good. local_path() is bad, local_path($path)
1264 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1265 already deprecated.
1267 Up to now remotebase was the string that described the remote root
1268 directory in rsync notation, like pause.perl.org::authors. And
1269 recentfile_basename was "RECENT-1h.yaml".
1271 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1273 * The test that was added in today's checkin is a good start for a test
1274 of rmirror. We should have more methods in Recent.pm: verify,
1275 addmissingfiles. We should verify the current tree, then rmirror it and
1276 then verifytree the copy. We could then add some arbitrary file and let
1277 it be discovered by addmissingfiles, then rmirror again and then
1278 verifytree the copy again.
1280 Then we could start stealing from csync2 sqlite database [no port to
1281 OSX!] and fill a local DB. And methods to compare the database with the
1282 recentfiles. Our strength is that in principle we could maintain state
1283 with a single float. We have synced up to 1234567890.123456. If the Z
1284 file does not add new files all we have to do is mirror the new ones and
1285 delete the goners.
1287 This makes it clear that we should extend current protocol and declare
1288 that we cheat when we add files too late, just to help the other end
1289 keeping track. Ah yes, that's what was meant when zloop was mentioned
1290 earlier.
1292 Maybe need to revisit File::Mirror to help me with this task.
1294 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1296 * There must be an allow-me-to-truncate flag in every recentfile.
1297 Without it one could construct a sequence of updates winning the locking
1298 battle against the aggregator. Only if an aggregator has managed to
1299 merge data over to the next level, truncating can be allowed. DONE with
1300 accessor merged.
1302 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1304 * We should probably guarantee that no duplicates enter the aggregator
1305 array.
1307 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1309 * To get merge operation faster would need a good benchmark test. What
1310 02 spits out isn't reliable enough and is dominated by many other
1311 things. Between
1313 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1314 Author: Andreas J. Koenig <k@k75.(none)>
1315 Date: Sat Aug 2 07:58:04 2008 +0200
1317 and
1319 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1320 Author: Andreas J. Koenig <k@k75.(none)>
1321 Date: Sat Aug 2 11:40:29 2008 +0200
1323 there was a lot of trying but the effect is hardly measurable with
1324 current tests.
1326 * overhead of connecting seems high. When setting
1327 max_files_per_connection to 1 we see that.
1329 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1331 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1333 25138.2215600014/3600 = 6.98283932222261
1335 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1336 be a bug in the aggregator. FIXED
1338 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1340 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1341 we have it, can we mirror it, mirror it!
1343 I fear this needs a new class which might be called
1344 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1345 and treat them as an entity. I realize that a single recentfile may be
1346 sufficient for certain tasks and that it is handy for the low level
1347 programmer but it is not nice to use. If there is a delete in the 1h
1348 file then the 6h file still contains it. Seekers of the best information
1349 need to combine at least some of the recentfiles most of the time.
1351 There is the place for the Z loop!
1353 But the combination is something to collect in a database, isn't it. Did
1354 csync2 just harrumph?
1356 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1358 * it just occurred to me that hosts in the same mirroring pool could
1359 help out each other even without rewriting the recentfile. Just fetch
1360 the stuff to mirror from several places, bingo. But that's something
1361 that should rather live in a separate package or in rsync directly.
1363 * cronjobs are unsuited because with ntp they would all come at the full
1364 minute and disturb each other. Besides that I'd hate to have a backbone
1365 with more than a few seconds latency.
1367 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1369 * a second rsync server with access control for PAUSE. Port? 873 is the
1370 standard port, let's take 8873.
1372 * if there were a filesystem based on this, it would have a slow access
1373 to inexistent files. It would probably provide wrong readdir (only based
1374 on current content) or also a slow one (based on a recentfile written
1375 after the call). But it would provide fast access to existing files. Or
1376 one would deliberately allow slightly blurred answers based on some
1377 sqlite reflection of the recentfiles.
1379 * todo: write a variant of mirror() that combines two or more
1380 recentfiles and treats them like one
1382 * todo: signal handler to remove the tempfile
1384 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1386 * now that we have the symlink I forgot how it should be used in
1387 practice.
1389 * the z loop: add missing files to Z file. Just append them (instead of
1390 prepending). So one guy prepends something from the Y file from time to
1391 time and another guy appends something rather frequently. Collecting
1392 pond. When Y merges into Z, things get epoch and the collecting pond
1393 gets smaller. What exactly are "missing files"?
1395 take note of current epoch of the alpha file, let's call it the
1396 recent-ts
1398 find all files on disk
1400 remove all files registered in the recentworld up to recent-ts
1402 remove all files that have been deleted after recent-ts according to
1403 recentworld
1405 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1407 * rersyncrecent might be a cronjob with a (locked) state file which
1408 contains things like after and maybe last z sync or such?
1410 rrr-mirror might be an alternative name but how would we justify the
1411 three Rs when there is no Re-Rsync-Recent?
1413 With the --loop parameter it is an endless loop, without it is no loop.
1414 At least this is simple.
1416 * todo: new accssor z-interval specifies how often the Z file is updated
1417 against the filesystem. We probably want no epoch stamp on these
1418 entries. And we want to be able to filter the entries (e.g. no
1419 by-modules and by-category tree)
1421 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1423 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1424 into Z. We do not need the other files filled up with filesystem
1425 contents though.
1427 * need interface to query for a file in order to NOT call update on
1428 PAUSE a second time within a short time.
1430 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1432 * recommended update interval? Makes no sense, is different for
1433 different users.
1435 * Moosify
1437 Local Variables:
1438 mode: change-log
1439 change-log-default-name: "Todo"
1440 tab-width: 2
1441 left-margin: 2
1442 End: