1 2009-05-02 Andreas J. Koenig <andk@cpan.org>
3 * Todo: rename rrr-dirtyupdate to rrr-update and add an option --dirty
4 so we can use it for normal update with a single file. And add a cronjob
5 that runs rrr-fsck so we get a reminder when something is not working
8 2009-04-27 Andreas J. Koenig <andk@cpan.org>
10 * Todo: I still see more link_stat errors than expected. I come back to
11 the former suspicion that we do not do necessary bookkeeping on deleted
12 files so that when we reach the "new" event on a meanwhile deleted file
13 we try to mirror it although we could know better. It's just noise, no
14 harm but noise is irritating.
16 * Todo: more experimenting with runstatusfile to get it official
18 * Todo: make sure $ENV{LANG}="C" when we call rsync because we will
21 rsync: link_stat "/authors/id/T/TH/THINC/DateTime-Format-Flexible-0.02.tar.gz" (in PAUSE) failed: No such file or directory
23 rsync: link_stat "/authors/id/A/AG/AGENT/SSH-Batch-0.001.meta" (in PAUSE) failed: No such file or directory (2)
25 And that ignore_link_stat_error is propagated in sparse_clone.
29 For the record: the bug was that ignore_link_stat_error was lost during
30 sparse_clone. And then the second bug was that it did not default to
31 true. Both are now fixed.
33 2009-04-25 Andreas J. Koenig <andk@cpan.org>
35 * Todo: fill rrr program with life
37 2009-04-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
39 * Minor bug: I see complaints about files not existing like
40 /authors/id/G/GW/GWADEJ/SVG-Sparkline-0.2.0.meta or
41 /authors/id/R/RC/RCAPUTO/POE-1.004.readme
43 In the RECENT files these are delete events but of course also new
44 events in older recent files.
46 I think we do not properly remember deletes when we skip-deletes and so
47 do not filter them out when we later see the "new" event.
49 Yes, but that what ignore_link_stat_error is for, we know about this
50 race condition. We must make ignore_link_stat_error default to true.
54 * bug in Done.pm failing to merge two fields into one when a third is
55 present. FIXED and accompanied with new test.
57 2009-04-24 Andreas J. Koenig <andk@cpan.org>
59 * Todo: keep the index clean and run some kind of nondestructive fsck 4
62 * who is our backbone?
64 http://cpan.cpantesters.org/authors/02STAMP # barbie
65 http://cpan.solfo.com/authors/02STAMP # abh
66 (http|ftp)://theoryx5.uwinnipeg.ca/pub/CPAN/ # rkobes
68 2009-04-17 Andreas J. Koenig <andk@cpan.org>
70 * the changelog helper I included in the Makefile release memo does not
71 show tags and is much less useful than I thought. gitk is probably much
74 2009-04-16 Andreas J. Koenig <andk@cpan.org>
78 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
79 #### Ydr_.yaml ... DONE
80 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
81 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
82 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
83 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
84 #### gone already at cpan-pause.pl line 0
86 Running without skip-deletes now but cannot reproduce.
88 2009-04-15 Andreas J. Koenig <andk@cpan.org>
90 * consider state directory such we can restart after a ^C or come again
93 Reuse _runstatusfile: write atomically, see how we can use the dump to
94 start running again. Do not leak and remove globs before writing. Do not
95 only dump one rf, dump the whole complex.
97 * consider the effect of ^C. Is it safe?
99 * want to use Perl::Version to up the versions on every release. Only if
100 things have changed, of course. Does this actually work?
102 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
103 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
105 Must do the bookkeeping myself when this should happen, maybe based on git?
107 I think I prefer the same version number for all pm files. But one also
108 needs a routine for setting them. Ahh:
110 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
116 cpan[2]> m /Mirror::Recent/
117 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
118 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
119 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
120 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
123 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
125 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
127 I just tried MM 6.5102 and it again produces only a provides for Recent,
128 not for the other modules.
130 My own fault. FIXED in the Makefile.PL.
132 2009-04-13 Andreas J. Koenig <andk@cpan.org>
134 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
136 * is it true (as stated at
137 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
138 that rsync can lead to errors when upstream changes before client sync
139 has completed? Can we see the error when we run this on upstream:
142 use Time::HiRes qw(time);
144 open my $fh, ">", "changingfile.txt.new" or die;
145 print $fh (time."\n") x 1000000;
147 rename "changingfile.txt.new", "changingfile.txt" or die;
151 And on the receiving end:
154 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
158 I cannot get it to fail. Apparently it is sufficient when upstream
159 always writes atomically (which of course is mandatory)?
161 SENT to mailing list Apr 28.
165 > https://fedorahosted.org/InstantMirror/
166 > https://www.redhat.com/mailman/listinfo/instantmirror-list
167 > irc.freenode.net channel #instantmirror
169 A cool name for a project. Inspires me to write a few sentences about
170 bittorrent's role in the grand picture.
172 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
173 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
175 2009-04-12 Andreas J. Koenig <andk@cpan.org>
177 * Interesting last minute bug during real download testing: the output
178 isn't as pretty anymore, it seems that more work is being done than
181 Introducing a highlevel _alluptodate method helped a lot in debugging
184 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
185 different on 1h file than on the other files. This lead to frequent
186 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
189 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
191 * Todo: try if we can be faster with a native float package. I'm really
192 glad to have a machine exposing perl floating point bugs, it forced me
193 to retract this item from the changes file: "several speedups in the
194 fakebigfloat code" before I get cpan testers fails.
196 2009-04-11 Andreas J. Koenig <andk@cpan.org>
198 * Bug: when the dirtymark changes upstream, then the downstream server
199 notices it immediately and mirrors the "1h" file but then leaves the
200 rmirror loop and continues with "6h" on the next round through. This
201 behaviour goes on until it reaches the Z file. Because we eat the Z file
202 piecemeal we leave the loop after a while but after that everybody seems
203 to have forgotten that there is some work left to be done. Bad, bad,
206 ETOOCONFUSING better name?
208 Recentfile::get_remote_recentfile_as_tempfile OK
209 Recentfile::resolve_recentfilename split_rfilename DONE
211 Recent::_principal_recentfile_object _principal_recentfile_fromremote
212 Recent::principal_recentfile OK
213 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
215 some void, some not, some not void but called in void context.
217 2009-04-10 Andreas J. Koenig <andk@cpan.org>
219 * Bug? When the dirtymark changes then the second tier hosts quickly
220 reset their DONE state and restart mirroring. But then they mirror
221 potentially outdated index files with inconsistent dirtymark because the
222 first tier box may be a bit behind on the large recentfiles.
224 This means we need a brake when iterating over the recentfiles that
225 refuses to use a recentfile with the wrong dirtymark.
227 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
229 * install the bin/* files and talk about rrr-overview somewhere.
231 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
232 created outside the target tree.
234 * Barbie asks for better logging capabilities
236 2009-03-29 Andreas J. Koenig <andk@cpan.org>
238 * known bug: cannot configure the Recentfile objects to keep delete
239 events. How to call the accessor for that? keep_delete_objects_forever?
241 Sounds acceptable. FIXED (but untested).
243 * known bug: keeps temporary index files lying around.
245 2009-03-24 Andreas J. Koenig <andk@cpan.org>
247 * Some equivalent for
249 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
253 2009-03-22 Andreas J. Koenig <andk@cpan.org>
255 * k81 is again client of k75 with different parameters.
257 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
258 about it all the time and nobody reacts accordingly; (2) lots of temp
259 files get created and not removed; culprit the new call to
260 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
261 caller has to remove the tempfile after use.
263 Need the drawing board.
265 PARTLY FIXED: retracting the idea to call
266 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
267 and unseed such that uptodate now means "we have mirrored this
268 recentfile's bunch of files" and seeded means "somebody believes the
269 index file for this rf needs be refreshed". This seems to work now.
271 Still collecting temp files that nobody cares to depollute.
273 2009-03-21 Andreas J. Koenig <andk@cpan.org>
275 * Bug?: should it be harder than it is atm to set the timestamp to the
278 * bug with native integers:
282 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
286 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
289 Native integer broke when native math was turned off. FIXED
291 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
292 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
295 Yes, records out of order:
298 447004 epoch: 995885533
299 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
302 447008 epoch: 995890358
303 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
306 447012 epoch: 995892221
307 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
310 FIXED with sanity check and later with the integer fix.
312 * Bug: want the index files in a .recent directory
314 * Bug: lots of dot files are not deleted in time
316 * possible test case: can a delete change the timestamp? This would
317 probably break the order of events.
319 2009-03-20 Andreas J. Koenig <andk@cpan.org>
321 * 1233701831.34486 what's so special about this number/string? It
323 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
324 atm lives in Y,Q, and Z.
326 It is the first entry after id/--skip-locking which has timestamp
327 1234164228.11325 which represents a file that doesn't exist anymore.
331 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
332 _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
333 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
334 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
335 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
336 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
337 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
338 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
339 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
340 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
341 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
342 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
343 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
345 FIXED, it was the "--skip-locking" file where manual intervention was participating
347 * bug on the mirroring slave: when the dirtymark gets increased we
348 probably do not reset the done intervals. The mirrorer stays within
349 tight bounds where it tries to sync with upstream and never seems to
350 finish. In the debugging state file I see lots of identical intervals
351 that do not get collapsed. When I restart the mirrorer it dies with:
353 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
354 _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
355 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
356 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
357 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
358 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
359 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
360 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
361 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
362 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
364 and debugging stands at
370 epoch: 1237400802.5789
371 path: id/J/JO/JOHND/CHECKSUMS
374 epoch: 1237400807.97514
375 path: id/M/MI/MIYAGAWA/CHECKSUMS
378 epoch: 1237400817.94363
379 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
387 and it is reproducable.
389 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
390 pause has a fresh one.
392 FIXED, it was the third anded term in each of the ifs in the IV block in
393 _register_one: with that we make sure that we do not stamp on valuable
396 2009-03-17 Andreas J. Koenig <andk@cpan.org>
398 * done: verified the existence of the floating point bug in bleadperl
399 and verified that switching from YAML::Syck to YAML::XS does not resolve
402 BTW, the switch was doable with
404 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
406 and should be considered as a separate TODO
408 * todo: integrate a dirty update with two aggregate calls before
409 unlocking for frictionless dirtying DONE
411 * todo: start the second rsync daemon on pause DONE
413 * todo: move index files to .recent: this cannot simply be done by
414 setting filenameroot to .recent/RECENT. Other parts of the modules rely
415 on the fact that dirname(recentfile) is the root of the mirrored tree.
417 2009-03-16 Andreas J. Koenig <andk@cpan.org>
419 * What was the resolution of the mirror.pl delete hook bug? Do we call
420 the delete hook when pause removes a file from MUIR?
422 * Today on pause: Updating 2a13fba..29f284d and installing it for
423 /usr/local/perl-5.10.0{,-RC2}
425 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
426 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
428 2009-03-15 Andreas J. Koenig <andk@cpan.org>
430 * currently recent_events has the side effect of setting dirtymark
431 because it forces a read on the file. That should be transparent, so
432 that the dirtymark call always forces a cache-able(?) read.
434 * The bug below is -- after a lot of trying -- not reproducible on a
435 small script, only in the large test script. The closest to the output
441 my $x = "01237123229.8814";
446 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
448 $l = "1237123231.22458";
449 $r = "1237123231.22458";
451 Devel::Peek::Dump $l;
452 Devel::Peek::Dump $r;
453 Devel::Peek::Dump $x = $l <=> $r;
457 The checked in state at c404a85 fails the test with my
458 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
459 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
460 in the test script always consistent.
462 * Todo: write a test that inserts a second dirty file with an already
463 existing timestamp. DONE
465 * Bug in perl 5.10 on my 64bit box:
467 DB<98> Devel::Peek::Dump $l
468 SV = PVMG(0x19e0450) at 0x142a550
470 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
472 NV = 1237123231.22458
473 PV = 0x194ce70 "1237123231.22458"\0
477 DB<99> Devel::Peek::Dump $r
478 SV = PVMG(0x19e0240) at 0x142a3e8
480 FLAGS = (PADMY,POK,pPOK)
483 PV = 0x19ff900 "1237123231.22458"\0
487 DB<100> Devel::Peek::Dump $l <=> $r
488 SV = IV(0x19ea6e8) at 0x19ea6f0
490 FLAGS = (PADTMP,IOK,pIOK)
493 DB<101> Devel::Peek::Dump $l
494 SV = PVMG(0x19e0450) at 0x142a550
496 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
498 NV = 1237123231.22458
499 PV = 0x194ce70 "1237123231.22458"\0
503 DB<102> Devel::Peek::Dump $r
504 SV = PVMG(0x19e0240) at 0x142a3e8
506 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
508 NV = 1237123231.22458
509 PV = 0x19ff900 "1237123231.22458"\0
513 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
515 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
517 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
519 * $obj->merge ($other) needs to learn about equal epoch which may happen
520 since dirty_epoch intruded.
522 * Wontfix anytime soon: I think we currently do not support mkdir. Only
525 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
527 * Todo: continue working on update(...,$dirty_epoch). It must be
528 followed by a fast_aggregate! DONE
530 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
532 * maybe we need a closest_entry or fitting_interval or something like
533 that. We want to merge an event into the middle of some recentfile.
534 First we do not know which file, then we do not know where to lock,
535 where to enter the new item, when and where to correct the dirtymark.
537 So my thought is we should first find which file.
539 Another part of my brain answers: what would happen if we would enter
540 the new file into the smallest file just like an ordinary new event,
541 just as an old event?
543 (1) we would write a duplicate timestamp? No, this would be easy to
546 (2) we would make the file large quickly? Yes, but so what? We are
547 changing the dirtymark, so are willing to disturb the downstream hosts.
549 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
551 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
555 https://rt.cpan.org/Ticket/Display.html?id=41199
557 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
559 Strangest output being HAKANARDO who managed to upload
561 Here is a better oneliner that includes also the first line of each
564 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
566 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
568 * memory leak in the syncher? It currently weighs 100M.
572 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
575 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
577 * lookup by epoch and by path and use this ability on the pause to never
578 again register a file twice that doesn't need it. Let's call it
581 * after the dirtymark is done: fill up recentfiles with fake (historic)
582 entries; fill up with individual corrections; algorithm maybe to be done
583 with bigfloat so that we can always place something in the middle
584 between two entries. Before we must switch to bigfloat we could try to
585 use Data::Float::nextup to get the.
587 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
590 * dirtymark now lives in Recentfile, needs to be used in rmirror.
592 * find out why the downloader died after a couple of hours without a net
593 connection. Write a test that survives the not-existence of the other
596 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
598 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
599 it makes less sense with tight coupling for secondary files. But asking
600 the server what the current epoch is might be cheaper on HTTP than on
601 rsync. (Needs to be evaluated)
603 * remove the 0.00 from the verbose overview in the Merged column in the
606 * write tests that expose the problems of the last few days: cascading
607 client/server roles, tight coupling for secondary RFs, deletes after
610 * Some day we might want to have policy options for the slave:
611 tight/loose/no coupling with upstream for secondary RFs. tight is what
612 we have now. loose would wait until a gap occurs that can be closed.
614 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
616 * revisit all $rfs->[$i+1] places if they now make sense still
618 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
620 * another bug is the fact that the mirror command deletes files before
621 it unhides the index file, thus confusing downstream slaves. We must not
622 delete before unhiding and must delete after unhiding. FIXED.
624 * new complication about the slave that is playing a server role.
625 Currently we mirror from newest to oldest with a hidden temporary file
626 as index. And when one file is finished, we unhide the index file.
627 Imagine the cascading server/slave is dead for a day. It then starts
628 mirroring again with the freshest thing and unhides the freshest index
629 file when it has worked through it. In that moment it exposes a time
630 hole. Because it now works on the second recentfile which is still
633 We currently do nothing special to converge after such a drop out. At
634 least not intentionally and robustly and thought through.
636 The algorithm we use to seed the next file needs quite a lot of more
637 robustness than it currently has. Something to do with looking at the
638 merged element of the next rf and when it has dropped off, we seed
639 immediately. And if it ramains dropped off, we seed again, of course.
641 Nope, looking from smaller to larger RFS we look at the merged element
642 of this RF and at the minmax/max element of the next RF. If that
643 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
646 And we need a public accessor seed and unseed or seeded. But not the mix
647 of public and private stuff that then is used behind the back.
649 And then the secondary* stuff must go.
651 And we must understand what the impact is on the DONE system. Can it go
652 unnoticed that there was a hole? And could the DONE system have decided
653 the hole is covered? This should be testable with three directories where
654 the middle stops working for a while. Done->merge is suspicious, we must
655 stop it from merging non-conflatable neighbors due to broken continuity.
659 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
661 * Slaven suggests to have the current epoch or the whole current
662 recentfile available from the HTTP server and take it away with
663 keepalive. This direction goes the granularity down to subseconds.
665 We might want to rewrite everything to factor out transport and allow
666 the whole thing to run via HTTP.
668 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
670 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
673 * smoker on k81 fetching from k75 to verify cascading works. See
674 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
676 * maybe the loop should wait for CHECKSUMS file after every upload. And
677 CPAN.pm needs to deal with timestamps in the future.
679 * do not forget the dirtymark!
681 Text: have a new flag on recentfiles with the meaning: if this
682 changes, you're required to run a full rsync over all the files. The
683 reason why we set it would probably be: some foul happened. we injected
684 files in arbitrary places or didn't inject them although they changed.
685 The content of the flag? Timestamp? The relation between the
686 recentfiles would have to be inheritance from the principal, because any
687 out of band changes would soon later propagate to the next recentfile.
689 By upping the flag often one can easily ruin the slaves.
691 last out of band change? dirtymark?
693 Anyway, this implies that we read a potentially existing recentfile
696 And it implies that we have an eventloop that keeps us busy in 2-3
697 cycles, one for current stuff (tight loop) and one for the recentfiles
698 (cascade when principal has changed), one for the old stuff after a
701 And it implies that the out-of-band change in any of the recentfiles
702 must have a lock on the principal file and there is the place to set the
705 * start a FAQ, especially quick start guide questions. Also to aid those
706 problematic areas where we have no good solution, like the "links"
709 * wish feedback when we are slow.
713 * Remove a few DEBUG statements.
715 * The multiple-rrr way of doing things needs a new option to rmirror,
716 like piecemeal or so. Not urgent because after the first pass through,
717 things run smoothely. It's only ugly during the first pass.
719 * I have the suspicion that the code is broken that decides if the
720 neighboring RF needs to be seeded. I fear when too much time has gone
721 between two calls (in our case more than one hour), it would not seed
722 the neighbor. Of course this will never be noticed, so we need a good
725 * local/localroot confusion: I currently pass both options but one must
728 * accounts for early birds on PAUSE rsync daemon.
730 * hardcoded 20 seconds
732 * who mirrors the index? DOING now.
734 * which CPAN mirrors offer rsync?
736 * visit all XXX, visit all _float places
738 * rename the pathdb stuff, it's too confusing. No idea how.
740 * rrr-inotify, backpan, rrr-register
742 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
744 * current bugs: the pathdb seems to get no reset, the seeding of the
745 secondaryttl stuff seems not to have an effect. Have helped myself with
746 a rand(10), need to fix this back. So not checked in. Does the rand
749 The rand thing helps. The secondaryttl stuff was in the wrong line,
752 The pathdb stuff was because I called either _pathdb or __pathdb on the
753 wrong object. FIXED now.
755 * It's not so beautiful if we never fetch the recentfiles that are not
756 the principal, even if this is correct behaviour. We really do not need
757 them after we have fetched the whole content.
759 OK, we want a switch for that: secondaryttl DONE
761 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
763 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
764 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
765 it lets 40 follow. FIXED
767 * See that the long running process really only updates the principal
768 file unless it has missed a timespan during which something happened. If
769 nothing happened, it must notice even when it misses the timespan. DONE
771 * we must throw away the pathdb when we have reached the end of Z. From
772 that moment we can have a very small pathdb because the only reason for
773 a pathdb is that we know to ignore old records in old files. We won't
774 need this pathdb again before the next full pass over the data is
775 necessary and then we will rebuild it as we go along. DONE
777 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
779 * I think, Done::register_one is doing wrong in that it does not
780 conflate neighboring pieces. The covered() method cannot do this because
781 it has no recent_events array at hand. But register_one has it and could
782 do it and for some reason misses to do it (sometimes).
784 This means that the three tests I just wrote can probably not survive
785 because they test with an already broken Done structure.
787 The art now is to detect how it happens, then to reproduce, then write a
790 So from the logfile this is what happens: we have a good interval with
791 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
792 top of it. Locally we now mirror F2 and open a new done interval for it.
793 Then we mirror F1 but this time with the timestamp T1b. And when we then
794 try to close the gap, we do not find T1 but instead something older. We
795 should gladly accept this older piece and this would fix this bug.
799 * bug to fix: when the 1h file changes while rmirror is running, we do
800 correctly sync the new files but never switch to the 6h file but rather
801 stay in a rather quick loop that fetches the 1h file again and again.
803 Is it possible that we initialize a new object? Or does
804 get_remote_recentfile_as_tempfile overwrite something in myself?
806 Want a new option: _runstatusfile => $file which frequently dumps the
807 state of all recentfiles to a file.
811 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
813 * Todo: now teach update to verify the timestamp is about to write
814 against the previous and use _increase_a_bit if it doesn't comply with
815 strict monotony. DONE
817 * The problem of rounding. So far perl's default precision was
818 sufficient. One day it won't be. FakeFloat has an easy job when it is
819 only reading and other machines have written correctly. But when we want
820 to write a floating point number that is a bit larger than the other
821 one, then we need our own idea of precision.
823 Slaven said: just append a "1". This might be going towards the end of
824 usability too quickly. I'd like something that actually uses the decimal
825 system. Well, appending a 1 also does this but...
827 E.g. we have 1.0. nextup on this architecture is starting with
828 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
829 taken the 1.0000000000000003 and the next user comes and the time tells
830 him 1.0 again. He has to beat my number without stepping over the
831 nextup. This is much less space than I had when I chose 1,2,3.
833 What is also irritating is that nextup is architecture dependent. The
834 128 bit guy must choose very long numbers to fit in between whereas the
835 other one with 16 bit uses larger steps. But then the algorithm is the
836 same for both, so that would be a nice thing.
838 I see two situation where we need this. One is when Time::HiRes returns
839 us a value that is <= the last entry in our recentfile. In this case
840 (let's call it the end-case) we must fill the region between that number
841 and the next higher native floating point number. The other is when we
842 inject an old file into an old recentfile (we would then also set a new
843 dirtymark). We find the integer value already taken and need a slightly
844 different one (let's call it the middle-case). The difference between
845 the two situations is that the next user will want to find something
846 higher than my number in the end-case and something lower than my number
849 So I suggest we give the function both a value and an upper bound and it
850 calculates us a primitive middle. The upper bound in the middle-case is
851 the next integer. The upper bound on the end-case is the nextup floating
852 point number. But the latter poses another problem: if we have occupied
853 the middle m between x and nextup(x), then the nextup(m) will probably
854 not be the same as nextup(x) because some rounding will take place
855 before the nextup is calculated and when the rounding reaches the
856 nextup(x), we will end up at nextup(nextup(x)).
858 So we really need to consider the nextup and the nextdown from there and
859 then the middle and that's the number we may approach asymptotically.
862 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
864 * consider deprecating the use of RECENT.recent as a symlink. It turns
865 out to need extra hoops with the rsync options and just isn't worth it.
866 Or maybe these extra hoops are needed anyway for the rest of the tree?
867 Nope, can't be the case because not all filesystems support symlinks.
869 But before doing the large step, I'll deprecate the call of
870 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
871 only introduced to resolve RECENT.recent and complicates the routine far
872 beyond what it deserves.
874 DONE. Won't deprecate RECENT.recent, just moved its handling to the
877 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
879 * I think it's a bug that the rsync_option links must be set to true in
880 order to support RECENT.recent and that nobody cares to set it
881 automatically. Similar for ignore_link_stat_errors.
883 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
885 * Todo: find all todos together and make a plan what is missing for a
888 - verifytree or something like that. fsck maybe.
890 - rersyncrecent, the script itself? What it do?
892 - a way to only mirror the recentfiles without mirroring the whole
893 remote system such that people can decide to mirror only partially see
894 also 2008-08-30. .shadow-xxx directory? this also needed for a
895 filesystem that is still incomplete and might need the mirrorfiles for
898 - long living objects that mirror again and again. Inject something
899 into ta, see how it goes over to tb.
901 - how do we continue filling up the DONE system when we use an object
902 for the second time? "fully covered" and "uptodate" or new terminology.
904 - overview called on the wrong file should be understandable
906 - the meta data field that must change when we fake something up so that
907 the downstream people know they have to re-fetch everything.
909 - how tolerant are we against missing files upstream? how do we keep
910 track? there are legitimate cases where we did read upstream index right
911 before a file got deleted there and then find that file as new and want
912 it. There are other cases that are not self healing and must be tracked
915 - how, exactly, do we have to deal with deletes? With rsync errors?
917 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
918 authors) failed: No such file or directory (2)
920 The file above is a delete in 1h and a new in file 1M and the
921 delete in the locally running rmirror did not get propagated to the 1M
922 object. Bug. And the consequence is a standstill.
924 It seems that a slave that works with a file below the principal needs
925 to merge things all the way up to get rid of later deletes. Or keep
926 track of all deletes and skip them later. So we need a trackdeletes.pm
927 similar to the done.pm?
929 see also 2008-08-20 about spurious deletes that really have no add
930 counterpart and yet they are not wrong.
932 - consider the effect when resyncing the recentfile takes longer than
933 the time per loop. Then we never rsync any file. We need to diagnose
934 that and force an increase of that loop time. But when we later are fast
935 enough again because the net has recovered, then we need to switch back
936 to original parameters. ERm, no, it's enough to keep syncing at least
937 one file before refetching an index file.
939 - remember to verify that no temp files are left lying around and the
942 - status file for not long running jobs that want to track upstream with
945 - revisit all XXX _float areas and study Sub::Exporter DONE
947 - persistent DB even though we just said we do not need it. Just for
948 extended capabilities and time savings when, for example, upstream
949 announces a reset and we get new recentfiles and could then limit
950 ourselves to a subset of files (those that have a changed epoch) in a
951 first pass and would only then do the loop to verify the rest. Or
954 * Todo: aggregate files should know their feed and finding the principal
955 should be done stepwise. (?)
957 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
958 left locks around. DONE
960 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
962 * maybe extend the _overview so that it always says if and where the
963 last file is in the next file and where the next event in the next rf
964 would lie. No, don't like this anymore. REJECT
966 * take the two new redundant tests out again, only the third must
969 * Todo: add a sanity check if the merged structure is really pointing to
970 a different rf and that this different rf is larger. DONE
972 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
974 * now test, if they are overlapping. And test if there is a file in the
975 next rf that would fit into this rf's interval.
977 1h 1222324012.8474 1222322541.7963 0.4086
978 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
979 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
980 1W 1222313218.3626 1221708477.5829 167.9835
982 I suspect that somebody writes a merged timestamp without having merged
983 and then somebody else relies on it.
985 If aggregate is running, the intervals must not be extravagated, if it
986 is not running, there must not be bounds, the total number of events in
987 the system must be counted and must be controlled throughout the tests.
988 That the test required the additional update was probably nonsense,
989 because aggregate can cut pieces too. FIXED & DONE
991 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
993 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
996 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
998 * Most apparent bug at the moment is that the recentfiles are fetched
999 too often. Only the principal should be fetched and if it has not
1000 changed, the others should not be refetched. ATM I must admit that I'm
1001 happy that we refetch more often than needed because I can more easily
1002 fix bugs while the thing is running.
1004 * Let's say, 1220474966.19501 is a timestamp of a file that is already
1005 done but the done system does not know about it. The reason for the
1006 failure is not known and we never reach the status uptodate because of
1007 this. We must get over it.
1009 Later it turns out that the origin server had a bug somewhere.
1010 1220474966.19042 came after 1220474966.19501. Or better: it was in the
1011 array of the recentfile one position above. The bug was my own.
1013 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1015 * There is the race condition where the server does a delete and the
1016 slave does not yet know and then tries to download it because he sees
1017 the new. So for this time window we must be more tolerant against
1018 failure. If we cannot download a file, we should just skip it and should
1019 not retry immediately. The whole system should discover the lost thing
1020 later. Keeping track with the DONE system should really be a no brainer.
1022 But there is something more: the whole filesystem is a database and the
1023 recentfiles are one possible representation of it. It's a pretty useful
1024 representation I think that's why I have implemented something around
1025 it. But for strictly local operation it has little value. For local
1026 operation we would much rather have a database. So we would enter every
1027 recentfile reading and every rsync operation and for every file the last
1028 state change and what it leads to. Then we would always ignore older
1029 records without the efforts involved with recentfiles.
1031 The database would have: path,recentepoch,rsyncedon,deletedon
1033 Oh well, not yet clear where this leads to.
1035 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1037 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
1038 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
1040 * statusfile, maybe only on demand, alone to have a sharp debugging
1041 tool. It is locked and all recentfiles dump themselves into it and we
1042 can build a viewer that lets us know where we stand and what's inside.
1044 * remember: only the principal recentfile needs expiration, all others
1045 shall be expired by principal if it discovers that something has move
1048 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1050 * Always check if we stringify to a higher value than in the entry
1053 * And in covered make an additional check if we would be able to see a
1054 numerical difference between the two numbers and if we can't then switch
1055 to a different, more expensive algorithm. Do not want to be caught by
1056 floating surprises. DONE
1058 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1060 * caching has several aspects here: we can cache the interval of the
1061 recentfile which only will change when the mtime of the file changes. We
1062 must re-mirror the recentfile when its ttl has expired. Does have_read
1063 tell you anything? It counts nothing at all. Only the mtime is
1064 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
1065 thing the have_mirrored because it is unrelated to the mtime.
1067 * Robustness of floating point calculations! I always thought that the
1068 string calculated by the origin server for the floating representation
1069 of the epoch time is just a string. When we convert it to a number and
1070 later back to a string, the other computer might come to a different
1071 conclusion. This must not happen, we want to preserve it under any
1072 circumstances. I will have to write tests with overlong sequences that
1073 get lost in arithmetic and must see if all still works well. DONE
1075 But one fragile point remains: if one host considers a>b and the other
1076 one considers them == but no eq. To prevent this, we must probably do
1077 some extra homework. DONE
1079 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1081 * the concept of tracking DONE needs an object per recentfile that has
1082 something like these methods:
1084 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1086 covered() register() covered()
1088 The unclear thing is how we translate points in time into intervals. We
1089 could pass a reference to the current recent_events array when running
1090 we_have(xxx) and let the DONE object iterate over it such that it only
1091 has to store a list of intervals that can melt into each other. Ah, even
1092 passing the list together with a list of indexes seems feasiable.
1094 Or maybe ask for the inverted list?
1096 Whenever the complete array is covered by the interval we say we are
1097 fully covered and if the recentfile is not expired, we are uptodate.
1099 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1101 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1103 * need a way to "return" the next entry after the end of a list. When
1104 the caller says "before" or "after" we would like to know if he could
1105 cover that interval/threshold or not because this influences the effect
1106 of a newer timestamp of that recentfile. DONE with $opt{info}.
1108 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1110 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1112 loop implies tixu (?). Nope, something like --statefile decides. Per
1115 T test, I init, X including removals, U nodirtymark
1117 So we have no concept of dirtymarks, we only trust that since we are
1118 running we have observed everything steadily. But people will not let
1119 this program run forever so we must consider both startup penalty and
1120 book keeping for later runs. We keep this for later. For now we write a
1121 long running mirror that merges several intervals.
1123 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1125 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1128 * rersyncrecent, the script: default to one week. The name of the switch
1129 is --after. Other switches? --loop!
1131 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1133 * need a switch --skip-deletes (?)
1135 * need a switch --enduser that tells us that the whole tempfile
1136 discipline is not needed when there is no downstream user. (?)
1138 Without this switch we cannot have a reasonable recent.pl that just
1139 displays the recent additions. Either we accept to download everything.
1140 Or we download temporary files without the typical rsync protocol
1143 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1144 File::Temp::tempdir, this might be a win.
1146 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1148 * apropos missing: we have no push, we never know the downstream
1149 servers. People who know their downstream hosts and want to ascertain
1150 something will want additional methods we have never thought about, like
1151 update or delete a certain file.
1153 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1155 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1156 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1159 * rsync options: it's a bit of a pain that we usually need several rsync
1160 options, like compress, links, times, checksum and that there is no
1161 reasonable default except the original rsync default. I think wee can
1162 safely assume that the rsync options are shared between all recentfile
1163 instances within one recent tree.
1165 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1167 * deletes: if a delete follows an add quickly enough it may happen that
1168 a downstream mirror did not see the add at all! It seems this needs to
1169 be mentioned somewhere. The point here is that even if the downstream is
1170 never missing the principal timeframe it may encounter a "delete" that
1171 has no complimentary "add" anywhere.
1173 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1175 * I suspect the treat of metadata is incorrect during read or something.
1176 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1177 contained more than 6 hours worth of data. At 06:08 we merged into the
1178 1d file. We need to take snapshots of the 6h file over the course of an
1179 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1182 Much worse: watching the 1h file: right at the moment (at 06:35) it
1183 covers 1218867584-1219120397 which is 70 hours.
1185 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1186 08:19:44 2008, that is when I checked out last time, so it seems to be
1187 aggregating and never truncating?
1189 No, correct is: it is never truncating; but wrong is: it is aggregating.
1190 It does receive a lot of events from time to time from a larger file.
1191 Somehow a large file gets merged into the small one and because the
1192 "meta/merged" attribute is missing, nobody is paying attention. I
1193 believe that I can fix this by making sure that metadata are honoured
1194 during read. DONE and test adjusted.
1196 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1198 * grand renaming plan
1200 remotebase => remoteroot to fit well with localroot DONE
1201 local_path() => localroot seems to me should already work DONE
1202 recentfile_basename => rfilename no need to stress it has no slash DONE
1204 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1205 kern? It would anyway need a deprecation cycle because it is an
1206 important constructor.
1208 * I like the portability that Data::Serializer brings us but the price
1209 is that some day we might find out that it is slowing us a bit. We'll
1212 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1214 * should we not enter the interval of the principal (or the interval of
1215 the merging file?) in every aggregated/merged file?
1217 * we should aim at a first release and give up on thinking about
1218 sanitizing stuff and zloop. Let's just admit that a full traditional
1219 rsync is the only available sanitizer ATM. Otherwise it's complicated
1220 stuff: sanitizing on the origin server, sanitizing on the slaves,
1221 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1222 get the basics out before this becomes a major cause for mess.
1224 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1226 * On OSes not supporting symlinks we expect that RECENT.recent contains
1227 the contents of the principal recentfile. Actually this is identical on
1228 systems supporting symlinks. Simple, what follows from that is that we
1229 need to keep the serializer in the metadata because we cannot read it
1230 from the filename, doesn't it? Of course not. It's a chicken and egg
1231 problem. This leaves us with the problem to actually parse the
1232 serialized data to find out in which format it is. So who can do the 4
1233 or 5 magics we wanted to support? File::LibMagic?
1235 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1237 * remotebase and recentfile_basename are ugly names. Now that we need a
1238 word for the shortest/principal/driving recentfile too we should do
1241 localroot is good. rfile is good. local_path() is bad, local_path($path)
1242 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1245 Up to now remotebase was the string that described the remote root
1246 directory in rsync notation, like pause.perl.org::authors. And
1247 recentfile_basename was "RECENT-1h.yaml".
1249 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1251 * The test that was added in today's checkin is a good start for a test
1252 of rmirror. We should have more methods in Recent.pm: verify,
1253 addmissingfiles. We should verify the current tree, then rmirror it and
1254 then verifytree the copy. We could then add some arbitrary file and let
1255 it be discovered by addmissingfiles, then rmirror again and then
1256 verifytree the copy again.
1258 Then we could start stealing from csync2 sqlite database [no port to
1259 OSX!] and fill a local DB. And methods to compare the database with the
1260 recentfiles. Our strength is that in principle we could maintain state
1261 with a single float. We have synced up to 1234567890.123456. If the Z
1262 file does not add new files all we have to do is mirror the new ones and
1265 This makes it clear that we should extend current protocol and declare
1266 that we cheat when we add files too late, just to help the other end
1267 keeping track. Ah yes, that's what was meant when zloop was mentioned
1270 Maybe need to revisit File::Mirror to help me with this task.
1272 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1274 * There must be an allow-me-to-truncate flag in every recentfile.
1275 Without it one could construct a sequence of updates winning the locking
1276 battle against the aggregator. Only if an aggregator has managed to
1277 merge data over to the next level, truncating can be allowed. DONE with
1280 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1282 * We should probably guarantee that no duplicates enter the aggregator
1285 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1287 * To get merge operation faster would need a good benchmark test. What
1288 02 spits out isn't reliable enough and is dominated by many other
1291 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1292 Author: Andreas J. Koenig <k@k75.(none)>
1293 Date: Sat Aug 2 07:58:04 2008 +0200
1297 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1298 Author: Andreas J. Koenig <k@k75.(none)>
1299 Date: Sat Aug 2 11:40:29 2008 +0200
1301 there was a lot of trying but the effect is hardly measurable with
1304 * overhead of connecting seems high. When setting
1305 max_files_per_connection to 1 we see that.
1307 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1309 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1311 25138.2215600014/3600 = 6.98283932222261
1313 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1314 be a bug in the aggregator. FIXED
1316 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1318 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1319 we have it, can we mirror it, mirror it!
1321 I fear this needs a new class which might be called
1322 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1323 and treat them as an entity. I realize that a single recentfile may be
1324 sufficient for certain tasks and that it is handy for the low level
1325 programmer but it is not nice to use. If there is a delete in the 1h
1326 file then the 6h file still contains it. Seekers of the best information
1327 need to combine at least some of the recentfiles most of the time.
1329 There is the place for the Z loop!
1331 But the combination is something to collect in a database, isn't it. Did
1332 csync2 just harrumph?
1334 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1336 * it just occurred to me that hosts in the same mirroring pool could
1337 help out each other even without rewriting the recentfile. Just fetch
1338 the stuff to mirror from several places, bingo. But that's something
1339 that should rather live in a separate package or in rsync directly.
1341 * cronjobs are unsuited because with ntp they would all come at the full
1342 minute and disturb each other. Besides that I'd hate to have a backbone
1343 with more than a few seconds latency.
1345 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1347 * a second rsync server with access control for PAUSE. Port? 873 is the
1348 standard port, let's take 8873.
1350 * if there were a filesystem based on this, it would have a slow access
1351 to inexistent files. It would probably provide wrong readdir (only based
1352 on current content) or also a slow one (based on a recentfile written
1353 after the call). But it would provide fast access to existing files. Or
1354 one would deliberately allow slightly blurred answers based on some
1355 sqlite reflection of the recentfiles.
1357 * todo: write a variant of mirror() that combines two or more
1358 recentfiles and treats them like one
1360 * todo: signal handler to remove the tempfile
1362 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1364 * now that we have the symlink I forgot how it should be used in
1367 * the z loop: add missing files to Z file. Just append them (instead of
1368 prepending). So one guy prepends something from the Y file from time to
1369 time and another guy appends something rather frequently. Collecting
1370 pond. When Y merges into Z, things get epoch and the collecting pond
1371 gets smaller. What exactly are "missing files"?
1373 take note of current epoch of the alpha file, let's call it the
1376 find all files on disk
1378 remove all files registered in the recentworld up to recent-ts
1380 remove all files that have been deleted after recent-ts according to
1383 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1385 * rersyncrecent might be a cronjob with a (locked) state file which
1386 contains things like after and maybe last z sync or such?
1388 rrr-mirror might be an alternative name but how would we justify the
1389 three Rs when there is no Re-Rsync-Recent?
1391 With the --loop parameter it is an endless loop, without it is no loop.
1392 At least this is simple.
1394 * todo: new accssor z-interval specifies how often the Z file is updated
1395 against the filesystem. We probably want no epoch stamp on these
1396 entries. And we want to be able to filter the entries (e.g. no
1397 by-modules and by-category tree)
1399 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1401 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1402 into Z. We do not need the other files filled up with filesystem
1405 * need interface to query for a file in order to NOT call update on
1406 PAUSE a second time within a short time.
1408 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1410 * recommended update interval? Makes no sense, is different for
1417 change-log-default-name: "Todo"