1 2011-02-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * IN_Q_OVERFLOW needs to trigger an fsck, likewise the entry into the
6 * speed: a large directory remove that is done by the kernel in 1
7 seconds triggers minutes of bookkeeping (07:56:42 - 08:01:10 for ~3000
8 files, all due to Schlemiehl again. Want to do some collecting of
9 pending ops before actually writing to the RECENT files. Lock!
11 2011-01-30 Andreas J. Koenig <andk@cpan.org>
13 * rrr-init and rrr-server need to be used such:
19 Stupid but no showstopper ATM.
21 2010-10-26 Andreas J. Koenig <andk@cpan.org>
23 * forking slave is done, needs more testing.
25 2010-07-06 Andreas J. Koenig <andk@cpan.org>
27 * The two next most important goals are to get a useable rrr-server and
28 a forking slave implementation that uses virtually no memory after the
29 first pass through. Does it need the status file?
31 * rrr-init now being fast we can now remove the alpha alerts. Done.
33 2010-06-16 Andreas J. Koenig <andk@cpan.org>
35 * rrr-init seems to be sssslllloooowwww. At 3:49 I started it for the
36 video tree. After twelve minutes the file has reached a size of 500k. It
37 is locking and rewriting the YAML after every file. Wasn't there an
38 unlocked variant? For "init" one would not expect locked correctness and
39 a rewrite after every item. Given that this tends to become slower with
40 every file it is really unacceptable.
42 Reminds me of rrr-dirtyupdate (the former name of rrr-update). It seems
43 the code was not dirty enough and that was the reason why it was so
44 slow. And rrr-update does not have a --dirty switch, so it looks like we
45 have no way to force dirty and fast operation yet but we need it for
48 Irritating the bootstrapping race: the order {(1) fill initial
49 array, (2) start server} leaves the need for an fsck to cover the time
50 between 1 and 2. Given that regular fsck is needed in any case the price
51 is not high but still doesn't give a warm feeling. Maybe starting the
52 server will by itself do an fsck?
54 2010-06-15 Andreas J. Koenig <andk@cpan.org>
56 * remove the wild alpha alerts and say, it is alpha but parts of it are
57 already working very well or so.
59 2009-12-10 Andreas J. Koenig <andk@cpan.org>
61 * Prominent request for statusfile to safe memory consumption. I'd like
62 to work backwards: get a verification program that verifies how well in
63 sync two nodes are without using frmr itself. Pretend that the forking
64 agent already works and can be interrupted with ^C. Then let it die at
65 random and restart at random and see the memory consumption and the
66 failure rate over time. Compare it with the curve we have now.
68 Of course we also invented the status file as a sharp debugging tool.
69 I'd like to see it in action. See also: F:R:M:Recent::thaw().
71 2009-06-22 Andreas J. Koenig <andk@cpan.org>
73 * Todo: rrr-init to generate some index files so that rrr-server can be
76 2009-05-02 Andreas J. Koenig <andk@cpan.org>
78 * DONE: rename rrr-dirtyupdate to rrr-update and add an option --dirty
79 so we can use it for normal update with a single file. And add a cronjob
80 that runs rrr-fsck so we get a reminder when something is not working
83 2009-04-27 Andreas J. Koenig <andk@cpan.org>
85 * Todo: I still see more link_stat errors than expected. I come back to
86 the former suspicion that we do not do necessary bookkeeping on deleted
87 files so that when we reach the "new" event on a meanwhile deleted file
88 we try to mirror it although we could know better. It's just noise, no
89 harm but noise is irritating.
91 * Todo: more experimenting with runstatusfile to get it official
93 * Todo: make sure $ENV{LANG}="C" when we call rsync because we will
96 rsync: link_stat "/authors/id/T/TH/THINC/DateTime-Format-Flexible-0.02.tar.gz" (in PAUSE) failed: No such file or directory
98 rsync: link_stat "/authors/id/A/AG/AGENT/SSH-Batch-0.001.meta" (in PAUSE) failed: No such file or directory (2)
100 And that ignore_link_stat_error is propagated in sparse_clone.
104 For the record: the bug was that ignore_link_stat_error was lost during
105 sparse_clone. And then the second bug was that it did not default to
106 true. Both are now fixed.
108 2009-04-25 Andreas J. Koenig <andk@cpan.org>
110 * Todo: fill rrr program with life
112 2009-04-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
114 * Minor bug: I see complaints about files not existing like
115 /authors/id/G/GW/GWADEJ/SVG-Sparkline-0.2.0.meta or
116 /authors/id/R/RC/RCAPUTO/POE-1.004.readme
118 In the RECENT files these are delete events but of course also new
119 events in older recent files.
121 I think we do not properly remember deletes when we skip-deletes and so
122 do not filter them out when we later see the "new" event.
124 Yes, but that what ignore_link_stat_error is for, we know about this
125 race condition. We must make ignore_link_stat_error default to true.
129 * bug in Done.pm failing to merge two fields into one when a third is
130 present. FIXED and accompanied with new test.
132 2009-04-24 Andreas J. Koenig <andk@cpan.org>
134 * Todo: keep the index clean and run some kind of nondestructive fsck 4
137 * who is our backbone?
139 http://cpan.cpantesters.org/authors/02STAMP # barbie
140 http://cpan.solfo.com/authors/02STAMP # abh
141 (http|ftp)://theoryx5.uwinnipeg.ca/pub/CPAN/ # rkobes
143 2009-04-17 Andreas J. Koenig <andk@cpan.org>
145 * the changelog helper I included in the Makefile release memo does not
146 show tags and is much less useful than I thought. gitk is probably much
149 2009-04-16 Andreas J. Koenig <andk@cpan.org>
153 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
154 #### Ydr_.yaml ... DONE
155 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
156 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
157 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
158 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
159 #### gone already at cpan-pause.pl line 0
161 Running without skip-deletes now but cannot reproduce.
163 2009-04-15 Andreas J. Koenig <andk@cpan.org>
165 * consider state directory such we can restart after a ^C or come again
168 Reuse _runstatusfile: write atomically, see how we can use the dump to
169 start running again. Do not leak and remove globs before writing. Do not
170 only dump one rf, dump the whole complex.
172 * consider the effect of ^C. Is it safe?
174 * want to use Perl::Version to up the versions on every release. Only if
175 things have changed, of course. Does this actually work?
177 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
178 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
180 Must do the bookkeeping myself when this should happen, maybe based on git?
182 I think I prefer the same version number for all pm files. But one also
183 needs a routine for setting them. Ahh:
185 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
191 cpan[2]> m /Mirror::Recent/
192 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
193 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
194 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
195 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
198 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
200 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
202 I just tried MM 6.5102 and it again produces only a provides for Recent,
203 not for the other modules.
205 My own fault. FIXED in the Makefile.PL.
207 2009-04-13 Andreas J. Koenig <andk@cpan.org>
209 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
211 * is it true (as stated at
212 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
213 that rsync can lead to errors when upstream changes before client sync
214 has completed? Can we see the error when we run this on upstream:
217 use Time::HiRes qw(time);
219 open my $fh, ">", "changingfile.txt.new" or die;
220 print $fh (time."\n") x 1000000;
222 rename "changingfile.txt.new", "changingfile.txt" or die;
226 And on the receiving end:
229 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
233 I cannot get it to fail. Apparently it is sufficient when upstream
234 always writes atomically (which of course is mandatory)?
236 SENT to mailing list Apr 28.
240 > https://fedorahosted.org/InstantMirror/
241 > https://www.redhat.com/mailman/listinfo/instantmirror-list
242 > irc.freenode.net channel #instantmirror
244 A cool name for a project. Inspires me to write a few sentences about
245 bittorrent's role in the grand picture.
247 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
248 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
250 2009-04-12 Andreas J. Koenig <andk@cpan.org>
252 * Interesting last minute bug during real download testing: the output
253 isn't as pretty anymore, it seems that more work is being done than
256 Introducing a highlevel _alluptodate method helped a lot in debugging
259 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
260 different on 1h file than on the other files. This lead to frequent
261 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
264 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
266 * Todo: try if we can be faster with a native float package. I'm really
267 glad to have a machine exposing perl floating point bugs, it forced me
268 to retract this item from the changes file: "several speedups in the
269 fakebigfloat code" before I get cpan testers fails.
271 2009-04-11 Andreas J. Koenig <andk@cpan.org>
273 * Bug: when the dirtymark changes upstream, then the downstream server
274 notices it immediately and mirrors the "1h" file but then leaves the
275 rmirror loop and continues with "6h" on the next round through. This
276 behaviour goes on until it reaches the Z file. Because we eat the Z file
277 piecemeal we leave the loop after a while but after that everybody seems
278 to have forgotten that there is some work left to be done. Bad, bad,
281 ETOOCONFUSING better name?
283 Recentfile::get_remote_recentfile_as_tempfile OK
284 Recentfile::resolve_recentfilename split_rfilename DONE
286 Recent::_principal_recentfile_object _principal_recentfile_fromremote
287 Recent::principal_recentfile OK
288 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
290 some void, some not, some not void but called in void context.
292 2009-04-10 Andreas J. Koenig <andk@cpan.org>
294 * Bug? When the dirtymark changes then the second tier hosts quickly
295 reset their DONE state and restart mirroring. But then they mirror
296 potentially outdated index files with inconsistent dirtymark because the
297 first tier box may be a bit behind on the large recentfiles.
299 This means we need a brake when iterating over the recentfiles that
300 refuses to use a recentfile with the wrong dirtymark.
302 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
304 * install the bin/* files and talk about rrr-overview somewhere.
306 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
307 created outside the target tree.
309 * Barbie asks for better logging capabilities
311 2009-03-29 Andreas J. Koenig <andk@cpan.org>
313 * known bug: cannot configure the Recentfile objects to keep delete
314 events. How to call the accessor for that? keep_delete_objects_forever?
316 Sounds acceptable. FIXED (but untested).
318 * known bug: keeps temporary index files lying around.
320 2009-03-24 Andreas J. Koenig <andk@cpan.org>
322 * Some equivalent for
324 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
328 2009-03-22 Andreas J. Koenig <andk@cpan.org>
330 * k81 is again client of k75 with different parameters.
332 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
333 about it all the time and nobody reacts accordingly; (2) lots of temp
334 files get created and not removed; culprit the new call to
335 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
336 caller has to remove the tempfile after use.
338 Need the drawing board.
340 PARTLY FIXED: retracting the idea to call
341 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
342 and unseed such that uptodate now means "we have mirrored this
343 recentfile's bunch of files" and seeded means "somebody believes the
344 index file for this rf needs be refreshed". This seems to work now.
346 Still collecting temp files that nobody cares to depollute.
348 2009-03-21 Andreas J. Koenig <andk@cpan.org>
350 * Bug?: should it be harder than it is atm to set the timestamp to the
353 * bug with native integers:
357 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
361 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
364 Native integer broke when native math was turned off. FIXED
366 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
367 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
370 Yes, records out of order:
373 447004 epoch: 995885533
374 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
377 447008 epoch: 995890358
378 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
381 447012 epoch: 995892221
382 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
385 FIXED with sanity check and later with the integer fix.
387 * Bug: want the index files in a .recent directory
389 * Bug: lots of dot files are not deleted in time
391 * possible test case: can a delete change the timestamp? This would
392 probably break the order of events.
394 2009-03-20 Andreas J. Koenig <andk@cpan.org>
396 * 1233701831.34486 what's so special about this number/string? It
398 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
399 atm lives in Y,Q, and Z.
401 It is the first entry after id/--skip-locking which has timestamp
402 1234164228.11325 which represents a file that doesn't exist anymore.
406 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
407 _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
408 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
409 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
410 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
411 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
412 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
413 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
414 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
415 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
416 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
417 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
418 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
420 FIXED, it was the "--skip-locking" file where manual intervention was participating
422 * bug on the mirroring slave: when the dirtymark gets increased we
423 probably do not reset the done intervals. The mirrorer stays within
424 tight bounds where it tries to sync with upstream and never seems to
425 finish. In the debugging state file I see lots of identical intervals
426 that do not get collapsed. When I restart the mirrorer it dies with:
428 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
429 _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
430 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
431 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
432 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
433 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
434 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
435 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
436 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
437 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
439 and debugging stands at
445 epoch: 1237400802.5789
446 path: id/J/JO/JOHND/CHECKSUMS
449 epoch: 1237400807.97514
450 path: id/M/MI/MIYAGAWA/CHECKSUMS
453 epoch: 1237400817.94363
454 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
462 and it is reproducable.
464 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
465 pause has a fresh one.
467 FIXED, it was the third anded term in each of the ifs in the IV block in
468 _register_one: with that we make sure that we do not stamp on valuable
471 2009-03-17 Andreas J. Koenig <andk@cpan.org>
473 * done: verified the existence of the floating point bug in bleadperl
474 and verified that switching from YAML::Syck to YAML::XS does not resolve
477 BTW, the switch was doable with
479 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
481 and should be considered as a separate TODO
483 * todo: integrate a dirty update with two aggregate calls before
484 unlocking for frictionless dirtying DONE
486 * todo: start the second rsync daemon on pause DONE
488 * todo: move index files to .recent: this cannot simply be done by
489 setting filenameroot to .recent/RECENT. Other parts of the modules rely
490 on the fact that dirname(recentfile) is the root of the mirrored tree.
492 2009-03-16 Andreas J. Koenig <andk@cpan.org>
494 * What was the resolution of the mirror.pl delete hook bug? Do we call
495 the delete hook when pause removes a file from MUIR?
497 * Today on pause: Updating 2a13fba..29f284d and installing it for
498 /usr/local/perl-5.10.0{,-RC2}
500 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
501 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
503 2009-03-15 Andreas J. Koenig <andk@cpan.org>
505 * currently recent_events has the side effect of setting dirtymark
506 because it forces a read on the file. That should be transparent, so
507 that the dirtymark call always forces a cache-able(?) read.
509 * The bug below is -- after a lot of trying -- not reproducible on a
510 small script, only in the large test script. The closest to the output
516 my $x = "01237123229.8814";
521 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
523 $l = "1237123231.22458";
524 $r = "1237123231.22458";
526 Devel::Peek::Dump $l;
527 Devel::Peek::Dump $r;
528 Devel::Peek::Dump $x = $l <=> $r;
532 The checked in state at c404a85 fails the test with my
533 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
534 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
535 in the test script always consistent.
537 * Todo: write a test that inserts a second dirty file with an already
538 existing timestamp. DONE
540 * Bug in perl 5.10 on my 64bit box:
542 DB<98> Devel::Peek::Dump $l
543 SV = PVMG(0x19e0450) at 0x142a550
545 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
547 NV = 1237123231.22458
548 PV = 0x194ce70 "1237123231.22458"\0
552 DB<99> Devel::Peek::Dump $r
553 SV = PVMG(0x19e0240) at 0x142a3e8
555 FLAGS = (PADMY,POK,pPOK)
558 PV = 0x19ff900 "1237123231.22458"\0
562 DB<100> Devel::Peek::Dump $l <=> $r
563 SV = IV(0x19ea6e8) at 0x19ea6f0
565 FLAGS = (PADTMP,IOK,pIOK)
568 DB<101> Devel::Peek::Dump $l
569 SV = PVMG(0x19e0450) at 0x142a550
571 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
573 NV = 1237123231.22458
574 PV = 0x194ce70 "1237123231.22458"\0
578 DB<102> Devel::Peek::Dump $r
579 SV = PVMG(0x19e0240) at 0x142a3e8
581 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
583 NV = 1237123231.22458
584 PV = 0x19ff900 "1237123231.22458"\0
588 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
590 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
592 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
594 * $obj->merge ($other) needs to learn about equal epoch which may happen
595 since dirty_epoch intruded.
597 * Wontfix anytime soon: I think we currently do not support mkdir. Only
600 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
602 * Todo: continue working on update(...,$dirty_epoch). It must be
603 followed by a fast_aggregate! DONE
605 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
607 * maybe we need a closest_entry or fitting_interval or something like
608 that. We want to merge an event into the middle of some recentfile.
609 First we do not know which file, then we do not know where to lock,
610 where to enter the new item, when and where to correct the dirtymark.
612 So my thought is we should first find which file.
614 Another part of my brain answers: what would happen if we would enter
615 the new file into the smallest file just like an ordinary new event,
616 just as an old event?
618 (1) we would write a duplicate timestamp? No, this would be easy to
621 (2) we would make the file large quickly? Yes, but so what? We are
622 changing the dirtymark, so are willing to disturb the downstream hosts.
624 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
626 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
630 https://rt.cpan.org/Ticket/Display.html?id=41199
632 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
634 Strangest output being HAKANARDO who managed to upload
636 Here is a better oneliner that includes also the first line of each
639 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
641 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
643 * memory leak in the syncher? It currently weighs 100M.
647 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
650 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
652 * lookup by epoch and by path and use this ability on the pause to never
653 again register a file twice that doesn't need it. Let's call it
656 * after the dirtymark is done: fill up recentfiles with fake (historic)
657 entries; fill up with individual corrections; algorithm maybe to be done
658 with bigfloat so that we can always place something in the middle
659 between two entries. Before we must switch to bigfloat we could try to
660 use Data::Float::nextup to get the.
662 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
665 * dirtymark now lives in Recentfile, needs to be used in rmirror.
667 * find out why the downloader died after a couple of hours without a net
668 connection. Write a test that survives the not-existence of the other
671 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
673 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
674 it makes less sense with tight coupling for secondary files. But asking
675 the server what the current epoch is might be cheaper on HTTP than on
676 rsync. (Needs to be evaluated)
678 * remove the 0.00 from the verbose overview in the Merged column in the
681 * write tests that expose the problems of the last few days: cascading
682 client/server roles, tight coupling for secondary RFs, deletes after
685 * Some day we might want to have policy options for the slave:
686 tight/loose/no coupling with upstream for secondary RFs. tight is what
687 we have now. loose would wait until a gap occurs that can be closed.
689 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
691 * revisit all $rfs->[$i+1] places if they now make sense still
693 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
695 * another bug is the fact that the mirror command deletes files before
696 it unhides the index file, thus confusing downstream slaves. We must not
697 delete before unhiding and must delete after unhiding. FIXED.
699 * new complication about the slave that is playing a server role.
700 Currently we mirror from newest to oldest with a hidden temporary file
701 as index. And when one file is finished, we unhide the index file.
702 Imagine the cascading server/slave is dead for a day. It then starts
703 mirroring again with the freshest thing and unhides the freshest index
704 file when it has worked through it. In that moment it exposes a time
705 hole. Because it now works on the second recentfile which is still
708 We currently do nothing special to converge after such a drop out. At
709 least not intentionally and robustly and thought through.
711 The algorithm we use to seed the next file needs quite a lot of more
712 robustness than it currently has. Something to do with looking at the
713 merged element of the next rf and when it has dropped off, we seed
714 immediately. And if it ramains dropped off, we seed again, of course.
716 Nope, looking from smaller to larger RFS we look at the merged element
717 of this RF and at the minmax/max element of the next RF. If that
718 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
721 And we need a public accessor seed and unseed or seeded. But not the mix
722 of public and private stuff that then is used behind the back.
724 And then the secondary* stuff must go.
726 And we must understand what the impact is on the DONE system. Can it go
727 unnoticed that there was a hole? And could the DONE system have decided
728 the hole is covered? This should be testable with three directories where
729 the middle stops working for a while. Done->merge is suspicious, we must
730 stop it from merging non-conflatable neighbors due to broken continuity.
734 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
736 * Slaven suggests to have the current epoch or the whole current
737 recentfile available from the HTTP server and take it away with
738 keepalive. This direction goes the granularity down to subseconds.
740 We might want to rewrite everything to factor out transport and allow
741 the whole thing to run via HTTP.
743 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
745 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
748 * smoker on k81 fetching from k75 to verify cascading works. See
749 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
751 * maybe the loop should wait for CHECKSUMS file after every upload. And
752 CPAN.pm needs to deal with timestamps in the future.
754 * do not forget the dirtymark!
756 Text: have a new flag on recentfiles with the meaning: if this
757 changes, you're required to run a full rsync over all the files. The
758 reason why we set it would probably be: some foul happened. we injected
759 files in arbitrary places or didn't inject them although they changed.
760 The content of the flag? Timestamp? The relation between the
761 recentfiles would have to be inheritance from the principal, because any
762 out of band changes would soon later propagate to the next recentfile.
764 By upping the flag often one can easily ruin the slaves.
766 last out of band change? dirtymark?
768 Anyway, this implies that we read a potentially existing recentfile
771 And it implies that we have an eventloop that keeps us busy in 2-3
772 cycles, one for current stuff (tight loop) and one for the recentfiles
773 (cascade when principal has changed), one for the old stuff after a
776 And it implies that the out-of-band change in any of the recentfiles
777 must have a lock on the principal file and there is the place to set the
780 * start a FAQ, especially quick start guide questions. Also to aid those
781 problematic areas where we have no good solution, like the "links"
784 * wish feedback when we are slow.
788 * Remove a few DEBUG statements.
790 * The multiple-rrr way of doing things needs a new option to rmirror,
791 like piecemeal or so. Not urgent because after the first pass through,
792 things run smoothely. It's only ugly during the first pass.
794 * I have the suspicion that the code is broken that decides if the
795 neighboring RF needs to be seeded. I fear when too much time has gone
796 between two calls (in our case more than one hour), it would not seed
797 the neighbor. Of course this will never be noticed, so we need a good
800 * local/localroot confusion: I currently pass both options but one must
803 * accounts for early birds on PAUSE rsync daemon.
805 * hardcoded 20 seconds
807 * who mirrors the index? DOING now.
809 * which CPAN mirrors offer rsync?
811 * visit all XXX, visit all _float places
813 * rename the pathdb stuff, it's too confusing. No idea how.
815 * rrr-inotify, backpan, rrr-register
817 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
819 * current bugs: the pathdb seems to get no reset, the seeding of the
820 secondaryttl stuff seems not to have an effect. Have helped myself with
821 a rand(10), need to fix this back. So not checked in. Does the rand
824 The rand thing helps. The secondaryttl stuff was in the wrong line,
827 The pathdb stuff was because I called either _pathdb or __pathdb on the
828 wrong object. FIXED now.
830 * It's not so beautiful if we never fetch the recentfiles that are not
831 the principal, even if this is correct behaviour. We really do not need
832 them after we have fetched the whole content.
834 OK, we want a switch for that: secondaryttl DONE
836 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
838 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
839 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
840 it lets 40 follow. FIXED
842 * See that the long running process really only updates the principal
843 file unless it has missed a timespan during which something happened. If
844 nothing happened, it must notice even when it misses the timespan. DONE
846 * we must throw away the pathdb when we have reached the end of Z. From
847 that moment we can have a very small pathdb because the only reason for
848 a pathdb is that we know to ignore old records in old files. We won't
849 need this pathdb again before the next full pass over the data is
850 necessary and then we will rebuild it as we go along. DONE
852 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
854 * I think, Done::register_one is doing wrong in that it does not
855 conflate neighboring pieces. The covered() method cannot do this because
856 it has no recent_events array at hand. But register_one has it and could
857 do it and for some reason misses to do it (sometimes).
859 This means that the three tests I just wrote can probably not survive
860 because they test with an already broken Done structure.
862 The art now is to detect how it happens, then to reproduce, then write a
865 So from the logfile this is what happens: we have a good interval with
866 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
867 top of it. Locally we now mirror F2 and open a new done interval for it.
868 Then we mirror F1 but this time with the timestamp T1b. And when we then
869 try to close the gap, we do not find T1 but instead something older. We
870 should gladly accept this older piece and this would fix this bug.
874 * bug to fix: when the 1h file changes while rmirror is running, we do
875 correctly sync the new files but never switch to the 6h file but rather
876 stay in a rather quick loop that fetches the 1h file again and again.
878 Is it possible that we initialize a new object? Or does
879 get_remote_recentfile_as_tempfile overwrite something in myself?
881 Want a new option: _runstatusfile => $file which frequently dumps the
882 state of all recentfiles to a file.
886 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
888 * Todo: now teach update to verify the timestamp is about to write
889 against the previous and use _increase_a_bit if it doesn't comply with
890 strict monotony. DONE
892 * The problem of rounding. So far perl's default precision was
893 sufficient. One day it won't be. FakeFloat has an easy job when it is
894 only reading and other machines have written correctly. But when we want
895 to write a floating point number that is a bit larger than the other
896 one, then we need our own idea of precision.
898 Slaven said: just append a "1". This might be going towards the end of
899 usability too quickly. I'd like something that actually uses the decimal
900 system. Well, appending a 1 also does this but...
902 E.g. we have 1.0. nextup on this architecture is starting with
903 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
904 taken the 1.0000000000000003 and the next user comes and the time tells
905 him 1.0 again. He has to beat my number without stepping over the
906 nextup. This is much less space than I had when I chose 1,2,3.
908 What is also irritating is that nextup is architecture dependent. The
909 128 bit guy must choose very long numbers to fit in between whereas the
910 other one with 16 bit uses larger steps. But then the algorithm is the
911 same for both, so that would be a nice thing.
913 I see two situation where we need this. One is when Time::HiRes returns
914 us a value that is <= the last entry in our recentfile. In this case
915 (let's call it the end-case) we must fill the region between that number
916 and the next higher native floating point number. The other is when we
917 inject an old file into an old recentfile (we would then also set a new
918 dirtymark). We find the integer value already taken and need a slightly
919 different one (let's call it the middle-case). The difference between
920 the two situations is that the next user will want to find something
921 higher than my number in the end-case and something lower than my number
924 So I suggest we give the function both a value and an upper bound and it
925 calculates us a primitive middle. The upper bound in the middle-case is
926 the next integer. The upper bound on the end-case is the nextup floating
927 point number. But the latter poses another problem: if we have occupied
928 the middle m between x and nextup(x), then the nextup(m) will probably
929 not be the same as nextup(x) because some rounding will take place
930 before the nextup is calculated and when the rounding reaches the
931 nextup(x), we will end up at nextup(nextup(x)).
933 So we really need to consider the nextup and the nextdown from there and
934 then the middle and that's the number we may approach asymptotically.
937 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
939 * consider deprecating the use of RECENT.recent as a symlink. It turns
940 out to need extra hoops with the rsync options and just isn't worth it.
941 Or maybe these extra hoops are needed anyway for the rest of the tree?
942 Nope, can't be the case because not all filesystems support symlinks.
944 But before doing the large step, I'll deprecate the call of
945 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
946 only introduced to resolve RECENT.recent and complicates the routine far
947 beyond what it deserves.
949 DONE. Won't deprecate RECENT.recent, just moved its handling to the
952 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
954 * I think it's a bug that the rsync_option links must be set to true in
955 order to support RECENT.recent and that nobody cares to set it
956 automatically. Similar for ignore_link_stat_errors.
958 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
960 * Todo: find all todos together and make a plan what is missing for a
963 - verifytree or something like that. fsck maybe.
965 - rersyncrecent, the script itself? What it do?
967 - a way to only mirror the recentfiles without mirroring the whole
968 remote system such that people can decide to mirror only partially see
969 also 2008-08-30. .shadow-xxx directory? this also needed for a
970 filesystem that is still incomplete and might need the mirrorfiles for
973 - long living objects that mirror again and again. Inject something
974 into ta, see how it goes over to tb.
976 - how do we continue filling up the DONE system when we use an object
977 for the second time? "fully covered" and "uptodate" or new terminology.
979 - overview called on the wrong file should be understandable
981 - the meta data field that must change when we fake something up so that
982 the downstream people know they have to re-fetch everything.
984 - how tolerant are we against missing files upstream? how do we keep
985 track? there are legitimate cases where we did read upstream index right
986 before a file got deleted there and then find that file as new and want
987 it. There are other cases that are not self healing and must be tracked
990 - how, exactly, do we have to deal with deletes? With rsync errors?
992 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
993 authors) failed: No such file or directory (2)
995 The file above is a delete in 1h and a new in file 1M and the
996 delete in the locally running rmirror did not get propagated to the 1M
997 object. Bug. And the consequence is a standstill.
999 It seems that a slave that works with a file below the principal needs
1000 to merge things all the way up to get rid of later deletes. Or keep
1001 track of all deletes and skip them later. So we need a trackdeletes.pm
1002 similar to the done.pm?
1004 see also 2008-08-20 about spurious deletes that really have no add
1005 counterpart and yet they are not wrong.
1007 - consider the effect when resyncing the recentfile takes longer than
1008 the time per loop. Then we never rsync any file. We need to diagnose
1009 that and force an increase of that loop time. But when we later are fast
1010 enough again because the net has recovered, then we need to switch back
1011 to original parameters. ERm, no, it's enough to keep syncing at least
1012 one file before refetching an index file.
1014 - remember to verify that no temp files are left lying around and the
1017 - status file for not long running jobs that want to track upstream with
1020 - revisit all XXX _float areas and study Sub::Exporter DONE
1022 - persistent DB even though we just said we do not need it. Just for
1023 extended capabilities and time savings when, for example, upstream
1024 announces a reset and we get new recentfiles and could then limit
1025 ourselves to a subset of files (those that have a changed epoch) in a
1026 first pass and would only then do the loop to verify the rest. Or
1029 * Todo: aggregate files should know their feed and finding the principal
1030 should be done stepwise. (?)
1032 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
1033 left locks around. DONE
1035 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1037 * maybe extend the _overview so that it always says if and where the
1038 last file is in the next file and where the next event in the next rf
1039 would lie. No, don't like this anymore. REJECT
1041 * take the two new redundant tests out again, only the third must
1044 * Todo: add a sanity check if the merged structure is really pointing to
1045 a different rf and that this different rf is larger. DONE
1047 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1049 * now test, if they are overlapping. And test if there is a file in the
1050 next rf that would fit into this rf's interval.
1052 1h 1222324012.8474 1222322541.7963 0.4086
1053 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
1054 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
1055 1W 1222313218.3626 1221708477.5829 167.9835
1057 I suspect that somebody writes a merged timestamp without having merged
1058 and then somebody else relies on it.
1060 If aggregate is running, the intervals must not be extravagated, if it
1061 is not running, there must not be bounds, the total number of events in
1062 the system must be counted and must be controlled throughout the tests.
1063 That the test required the additional update was probably nonsense,
1064 because aggregate can cut pieces too. FIXED & DONE
1066 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1068 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
1071 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1073 * Most apparent bug at the moment is that the recentfiles are fetched
1074 too often. Only the principal should be fetched and if it has not
1075 changed, the others should not be refetched. ATM I must admit that I'm
1076 happy that we refetch more often than needed because I can more easily
1077 fix bugs while the thing is running.
1079 * Let's say, 1220474966.19501 is a timestamp of a file that is already
1080 done but the done system does not know about it. The reason for the
1081 failure is not known and we never reach the status uptodate because of
1082 this. We must get over it.
1084 Later it turns out that the origin server had a bug somewhere.
1085 1220474966.19042 came after 1220474966.19501. Or better: it was in the
1086 array of the recentfile one position above. The bug was my own.
1088 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1090 * There is the race condition where the server does a delete and the
1091 slave does not yet know and then tries to download it because he sees
1092 the new. So for this time window we must be more tolerant against
1093 failure. If we cannot download a file, we should just skip it and should
1094 not retry immediately. The whole system should discover the lost thing
1095 later. Keeping track with the DONE system should really be a no brainer.
1097 But there is something more: the whole filesystem is a database and the
1098 recentfiles are one possible representation of it. It's a pretty useful
1099 representation I think that's why I have implemented something around
1100 it. But for strictly local operation it has little value. For local
1101 operation we would much rather have a database. So we would enter every
1102 recentfile reading and every rsync operation and for every file the last
1103 state change and what it leads to. Then we would always ignore older
1104 records without the efforts involved with recentfiles.
1106 The database would have: path,recentepoch,rsyncedon,deletedon
1108 Oh well, not yet clear where this leads to.
1110 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1112 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
1113 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
1115 * statusfile, maybe only on demand, alone to have a sharp debugging
1116 tool. It is locked and all recentfiles dump themselves into it and we
1117 can build a viewer that lets us know where we stand and what's inside.
1119 * remember: only the principal recentfile needs expiration, all others
1120 shall be expired by principal if it discovers that something has move
1123 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1125 * Always check if we stringify to a higher value than in the entry
1128 * And in covered make an additional check if we would be able to see a
1129 numerical difference between the two numbers and if we can't then switch
1130 to a different, more expensive algorithm. Do not want to be caught by
1131 floating surprises. DONE
1133 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1135 * caching has several aspects here: we can cache the interval of the
1136 recentfile which only will change when the mtime of the file changes. We
1137 must re-mirror the recentfile when its ttl has expired. Does have_read
1138 tell you anything? It counts nothing at all. Only the mtime is
1139 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
1140 thing the have_mirrored because it is unrelated to the mtime.
1142 * Robustness of floating point calculations! I always thought that the
1143 string calculated by the origin server for the floating representation
1144 of the epoch time is just a string. When we convert it to a number and
1145 later back to a string, the other computer might come to a different
1146 conclusion. This must not happen, we want to preserve it under any
1147 circumstances. I will have to write tests with overlong sequences that
1148 get lost in arithmetic and must see if all still works well. DONE
1150 But one fragile point remains: if one host considers a>b and the other
1151 one considers them == but no eq. To prevent this, we must probably do
1152 some extra homework. DONE
1154 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1156 * the concept of tracking DONE needs an object per recentfile that has
1157 something like these methods:
1159 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1161 covered() register() covered()
1163 The unclear thing is how we translate points in time into intervals. We
1164 could pass a reference to the current recent_events array when running
1165 we_have(xxx) and let the DONE object iterate over it such that it only
1166 has to store a list of intervals that can melt into each other. Ah, even
1167 passing the list together with a list of indexes seems feasiable.
1169 Or maybe ask for the inverted list?
1171 Whenever the complete array is covered by the interval we say we are
1172 fully covered and if the recentfile is not expired, we are uptodate.
1174 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1176 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1178 * need a way to "return" the next entry after the end of a list. When
1179 the caller says "before" or "after" we would like to know if he could
1180 cover that interval/threshold or not because this influences the effect
1181 of a newer timestamp of that recentfile. DONE with $opt{info}.
1183 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1185 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1187 loop implies tixu (?). Nope, something like --statefile decides. Per
1190 T test, I init, X including removals, U nodirtymark
1192 So we have no concept of dirtymarks, we only trust that since we are
1193 running we have observed everything steadily. But people will not let
1194 this program run forever so we must consider both startup penalty and
1195 book keeping for later runs. We keep this for later. For now we write a
1196 long running mirror that merges several intervals.
1198 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1200 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1203 * rersyncrecent, the script: default to one week. The name of the switch
1204 is --after. Other switches? --loop!
1206 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1208 * need a switch --skip-deletes (?)
1210 * need a switch --enduser that tells us that the whole tempfile
1211 discipline is not needed when there is no downstream user. (?)
1213 Without this switch we cannot have a reasonable recent.pl that just
1214 displays the recent additions. Either we accept to download everything.
1215 Or we download temporary files without the typical rsync protocol
1218 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1219 File::Temp::tempdir, this might be a win.
1221 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1223 * apropos missing: we have no push, we never know the downstream
1224 servers. People who know their downstream hosts and want to ascertain
1225 something will want additional methods we have never thought about, like
1226 update or delete a certain file.
1228 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1230 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1231 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1234 * rsync options: it's a bit of a pain that we usually need several rsync
1235 options, like compress, links, times, checksum and that there is no
1236 reasonable default except the original rsync default. I think wee can
1237 safely assume that the rsync options are shared between all recentfile
1238 instances within one recent tree.
1240 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1242 * deletes: if a delete follows an add quickly enough it may happen that
1243 a downstream mirror did not see the add at all! It seems this needs to
1244 be mentioned somewhere. The point here is that even if the downstream is
1245 never missing the principal timeframe it may encounter a "delete" that
1246 has no complimentary "add" anywhere.
1248 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1250 * I suspect the treat of metadata is incorrect during read or something.
1251 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1252 contained more than 6 hours worth of data. At 06:08 we merged into the
1253 1d file. We need to take snapshots of the 6h file over the course of an
1254 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1257 Much worse: watching the 1h file: right at the moment (at 06:35) it
1258 covers 1218867584-1219120397 which is 70 hours.
1260 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1261 08:19:44 2008, that is when I checked out last time, so it seems to be
1262 aggregating and never truncating?
1264 No, correct is: it is never truncating; but wrong is: it is aggregating.
1265 It does receive a lot of events from time to time from a larger file.
1266 Somehow a large file gets merged into the small one and because the
1267 "meta/merged" attribute is missing, nobody is paying attention. I
1268 believe that I can fix this by making sure that metadata are honoured
1269 during read. DONE and test adjusted.
1271 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1273 * grand renaming plan
1275 remotebase => remoteroot to fit well with localroot DONE
1276 local_path() => localroot seems to me should already work DONE
1277 recentfile_basename => rfilename no need to stress it has no slash DONE
1279 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1280 kern? It would anyway need a deprecation cycle because it is an
1281 important constructor.
1283 * I like the portability that Data::Serializer brings us but the price
1284 is that some day we might find out that it is slowing us a bit. We'll
1287 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1289 * should we not enter the interval of the principal (or the interval of
1290 the merging file?) in every aggregated/merged file?
1292 * we should aim at a first release and give up on thinking about
1293 sanitizing stuff and zloop. Let's just admit that a full traditional
1294 rsync is the only available sanitizer ATM. Otherwise it's complicated
1295 stuff: sanitizing on the origin server, sanitizing on the slaves,
1296 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1297 get the basics out before this becomes a major cause for mess.
1299 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1301 * On OSes not supporting symlinks we expect that RECENT.recent contains
1302 the contents of the principal recentfile. Actually this is identical on
1303 systems supporting symlinks. Simple, what follows from that is that we
1304 need to keep the serializer in the metadata because we cannot read it
1305 from the filename, doesn't it? Of course not. It's a chicken and egg
1306 problem. This leaves us with the problem to actually parse the
1307 serialized data to find out in which format it is. So who can do the 4
1308 or 5 magics we wanted to support? File::LibMagic?
1310 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1312 * remotebase and recentfile_basename are ugly names. Now that we need a
1313 word for the shortest/principal/driving recentfile too we should do
1316 localroot is good. rfile is good. local_path() is bad, local_path($path)
1317 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1320 Up to now remotebase was the string that described the remote root
1321 directory in rsync notation, like pause.perl.org::authors. And
1322 recentfile_basename was "RECENT-1h.yaml".
1324 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1326 * The test that was added in today's checkin is a good start for a test
1327 of rmirror. We should have more methods in Recent.pm: verify,
1328 addmissingfiles. We should verify the current tree, then rmirror it and
1329 then verifytree the copy. We could then add some arbitrary file and let
1330 it be discovered by addmissingfiles, then rmirror again and then
1331 verifytree the copy again.
1333 Then we could start stealing from csync2 sqlite database [no port to
1334 OSX!] and fill a local DB. And methods to compare the database with the
1335 recentfiles. Our strength is that in principle we could maintain state
1336 with a single float. We have synced up to 1234567890.123456. If the Z
1337 file does not add new files all we have to do is mirror the new ones and
1340 This makes it clear that we should extend current protocol and declare
1341 that we cheat when we add files too late, just to help the other end
1342 keeping track. Ah yes, that's what was meant when zloop was mentioned
1345 Maybe need to revisit File::Mirror to help me with this task.
1347 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1349 * There must be an allow-me-to-truncate flag in every recentfile.
1350 Without it one could construct a sequence of updates winning the locking
1351 battle against the aggregator. Only if an aggregator has managed to
1352 merge data over to the next level, truncating can be allowed. DONE with
1355 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1357 * We should probably guarantee that no duplicates enter the aggregator
1360 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1362 * To get merge operation faster would need a good benchmark test. What
1363 02 spits out isn't reliable enough and is dominated by many other
1366 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1367 Author: Andreas J. Koenig <k@k75.(none)>
1368 Date: Sat Aug 2 07:58:04 2008 +0200
1372 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1373 Author: Andreas J. Koenig <k@k75.(none)>
1374 Date: Sat Aug 2 11:40:29 2008 +0200
1376 there was a lot of trying but the effect is hardly measurable with
1379 * overhead of connecting seems high. When setting
1380 max_files_per_connection to 1 we see that.
1382 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1384 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1386 25138.2215600014/3600 = 6.98283932222261
1388 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1389 be a bug in the aggregator. FIXED
1391 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1393 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1394 we have it, can we mirror it, mirror it!
1396 I fear this needs a new class which might be called
1397 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1398 and treat them as an entity. I realize that a single recentfile may be
1399 sufficient for certain tasks and that it is handy for the low level
1400 programmer but it is not nice to use. If there is a delete in the 1h
1401 file then the 6h file still contains it. Seekers of the best information
1402 need to combine at least some of the recentfiles most of the time.
1404 There is the place for the Z loop!
1406 But the combination is something to collect in a database, isn't it. Did
1407 csync2 just harrumph?
1409 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1411 * it just occurred to me that hosts in the same mirroring pool could
1412 help out each other even without rewriting the recentfile. Just fetch
1413 the stuff to mirror from several places, bingo. But that's something
1414 that should rather live in a separate package or in rsync directly.
1416 * cronjobs are unsuited because with ntp they would all come at the full
1417 minute and disturb each other. Besides that I'd hate to have a backbone
1418 with more than a few seconds latency.
1420 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1422 * a second rsync server with access control for PAUSE. Port? 873 is the
1423 standard port, let's take 8873.
1425 * if there were a filesystem based on this, it would have a slow access
1426 to inexistent files. It would probably provide wrong readdir (only based
1427 on current content) or also a slow one (based on a recentfile written
1428 after the call). But it would provide fast access to existing files. Or
1429 one would deliberately allow slightly blurred answers based on some
1430 sqlite reflection of the recentfiles.
1432 * todo: write a variant of mirror() that combines two or more
1433 recentfiles and treats them like one
1435 * todo: signal handler to remove the tempfile
1437 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1439 * now that we have the symlink I forgot how it should be used in
1442 * the z loop: add missing files to Z file. Just append them (instead of
1443 prepending). So one guy prepends something from the Y file from time to
1444 time and another guy appends something rather frequently. Collecting
1445 pond. When Y merges into Z, things get epoch and the collecting pond
1446 gets smaller. What exactly are "missing files"?
1448 take note of current epoch of the alpha file, let's call it the
1451 find all files on disk
1453 remove all files registered in the recentworld up to recent-ts
1455 remove all files that have been deleted after recent-ts according to
1458 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1460 * rersyncrecent might be a cronjob with a (locked) state file which
1461 contains things like after and maybe last z sync or such?
1463 rrr-mirror might be an alternative name but how would we justify the
1464 three Rs when there is no Re-Rsync-Recent?
1466 With the --loop parameter it is an endless loop, without it is no loop.
1467 At least this is simple.
1469 * todo: new accssor z-interval specifies how often the Z file is updated
1470 against the filesystem. We probably want no epoch stamp on these
1471 entries. And we want to be able to filter the entries (e.g. no
1472 by-modules and by-category tree)
1474 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1476 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1477 into Z. We do not need the other files filled up with filesystem
1480 * need interface to query for a file in order to NOT call update on
1481 PAUSE a second time within a short time.
1483 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1485 * recommended update interval? Makes no sense, is different for
1492 change-log-default-name: "Todo"