1 2009-04-27 Andreas J. Koenig <andk@cpan.org>
3 * Todo: I still see more link_stat errors than expected. I come back to
4 the former suspicion that we do not do necessary bookkeeping on deleted
5 files so that when we reach the "new" event on a meanwhile deleted file
6 we try to mirror it although we could know better. It's just noise, no
7 harm but noise is irritating.
9 * Todo: more experimenting with runstatusfile to get it official
11 * Todo: make sure $ENV{LANG}="C" when we call rsync because we will
14 rsync: link_stat "/authors/id/T/TH/THINC/DateTime-Format-Flexible-0.02.tar.gz" (in PAUSE) failed: No such file or directory
16 rsync: link_stat "/authors/id/A/AG/AGENT/SSH-Batch-0.001.meta" (in PAUSE) failed: No such file or directory (2)
18 And that ignore_link_stat_error is propagated in sparse_clone.
22 For the record: the bug was that ignore_link_stat_error was lost during
23 sparse_clone. And then the second bug was that it did not default to
24 true. Both are now fixed.
26 2009-04-25 Andreas J. Koenig <andk@cpan.org>
28 * Todo: fill rrr program with life
30 2009-04-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
32 * Minor bug: I see complaints about files not existing like
33 /authors/id/G/GW/GWADEJ/SVG-Sparkline-0.2.0.meta or
34 /authors/id/R/RC/RCAPUTO/POE-1.004.readme
36 In the RECENT files these are delete events but of course also new
37 events in older recent files.
39 I think we do not properly remember deletes when we skip-deletes and so
40 do not filter them out when we later see the "new" event.
42 Yes, but that what ignore_link_stat_error is for, we know about this
43 race condition. We must make ignore_link_stat_error default to true.
47 * bug in Done.pm failing to merge two fields into one when a third is
48 present. FIXED and accompanied with new test.
50 2009-04-24 Andreas J. Koenig <andk@cpan.org>
52 * Todo: keep the index clean and run some kind of nondestructive fsck 4
55 * who is our backbone?
57 http://cpan.cpantesters.org/authors/02STAMP # barbie
58 http://cpan.solfo.com/authors/02STAMP # abh
59 (http|ftp)://theoryx5.uwinnipeg.ca/pub/CPAN/ # rkobes
61 2009-04-17 Andreas J. Koenig <andk@cpan.org>
63 * the changelog helper I included in the Makefile release memo does not
64 show tags and is much less useful than I thought. gitk is probably much
67 2009-04-16 Andreas J. Koenig <andk@cpan.org>
71 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
72 #### Ydr_.yaml ... DONE
73 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
74 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
75 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
76 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
77 #### gone already at cpan-pause.pl line 0
79 Running without skip-deletes now but cannot reproduce.
81 2009-04-15 Andreas J. Koenig <andk@cpan.org>
83 * consider state directory such we can restart after a ^C or come again
86 Reuse _runstatusfile: write atomically, see how we can use the dump to
87 start running again. Do not leak and remove globs before writing. Do not
88 only dump one rf, dump the whole complex.
90 * consider the effect of ^C. Is it safe?
92 * want to use Perl::Version to up the versions on every release. Only if
93 things have changed, of course. Does this actually work?
95 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
96 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
98 Must do the bookkeeping myself when this should happen, maybe based on git?
100 I think I prefer the same version number for all pm files. But one also
101 needs a routine for setting them. Ahh:
103 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
109 cpan[2]> m /Mirror::Recent/
110 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
111 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
112 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
113 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
116 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
118 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
120 I just tried MM 6.5102 and it again produces only a provides for Recent,
121 not for the other modules.
123 My own fault. FIXED in the Makefile.PL.
125 2009-04-13 Andreas J. Koenig <andk@cpan.org>
127 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
129 * is it true (as stated at
130 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
131 that rsync can lead to errors when upstream changes before client sync
132 has completed? Can we see the error when we run this on upstream:
135 use Time::HiRes qw(time);
137 open my $fh, ">", "changingfile.txt.new" or die;
138 print $fh (time."\n") x 1000000;
140 rename "changingfile.txt.new", "changingfile.txt" or die;
144 And on the receiving end:
147 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
151 I cannot get it to fail. Apparently it is sufficient when upstream
152 always writes atomically (which of course is mandatory)?
154 SENT to mailing list Apr 28.
158 > https://fedorahosted.org/InstantMirror/
159 > https://www.redhat.com/mailman/listinfo/instantmirror-list
160 > irc.freenode.net channel #instantmirror
162 A cool name for a project. Inspires me to write a few sentences about
163 bittorrent's role in the grand picture.
165 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
166 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
168 2009-04-12 Andreas J. Koenig <andk@cpan.org>
170 * Interesting last minute bug during real download testing: the output
171 isn't as pretty anymore, it seems that more work is being done than
174 Introducing a highlevel _alluptodate method helped a lot in debugging
177 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
178 different on 1h file than on the other files. This lead to frequent
179 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
182 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
184 * Todo: try if we can be faster with a native float package. I'm really
185 glad to have a machine exposing perl floating point bugs, it forced me
186 to retract this item from the changes file: "several speedups in the
187 fakebigfloat code" before I get cpan testers fails.
189 2009-04-11 Andreas J. Koenig <andk@cpan.org>
191 * Bug: when the dirtymark changes upstream, then the downstream server
192 notices it immediately and mirrors the "1h" file but then leaves the
193 rmirror loop and continues with "6h" on the next round through. This
194 behaviour goes on until it reaches the Z file. Because we eat the Z file
195 piecemeal we leave the loop after a while but after that everybody seems
196 to have forgotten that there is some work left to be done. Bad, bad,
199 ETOOCONFUSING better name?
201 Recentfile::get_remote_recentfile_as_tempfile OK
202 Recentfile::resolve_recentfilename split_rfilename DONE
204 Recent::_principal_recentfile_object _principal_recentfile_fromremote
205 Recent::principal_recentfile OK
206 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
208 some void, some not, some not void but called in void context.
210 2009-04-10 Andreas J. Koenig <andk@cpan.org>
212 * Todo: rename rrr-dirtyupdate to rrr-update and add an option --dirty
213 so we can use it for normal update with a single file.
215 * Bug? When the dirtymark changes then the second tier hosts quickly
216 reset their DONE state and restart mirroring. But then they mirror
217 potentially outdated index files with inconsistent dirtymark because the
218 first tier box may be a bit behind on the large recentfiles.
220 This means we need a brake when iterating over the recentfiles that
221 refuses to use a recentfile with the wrong dirtymark.
223 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
225 * install the bin/* files and talk about rrr-overview somewhere.
227 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
228 created outside the target tree.
230 * Barbie asks for better logging capabilities
232 2009-03-29 Andreas J. Koenig <andk@cpan.org>
234 * known bug: cannot configure the Recentfile objects to keep delete
235 events. How to call the accessor for that? keep_delete_objects_forever?
237 Sounds acceptable. FIXED (but untested).
239 * known bug: keeps temporary index files lying around.
241 2009-03-24 Andreas J. Koenig <andk@cpan.org>
243 * Some equivalent for
245 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
249 2009-03-22 Andreas J. Koenig <andk@cpan.org>
251 * k81 is again client of k75 with different parameters.
253 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
254 about it all the time and nobody reacts accordingly; (2) lots of temp
255 files get created and not removed; culprit the new call to
256 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
257 caller has to remove the tempfile after use.
259 Need the drawing board.
261 PARTLY FIXED: retracting the idea to call
262 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
263 and unseed such that uptodate now means "we have mirrored this
264 recentfile's bunch of files" and seeded means "somebody believes the
265 index file for this rf needs be refreshed". This seems to work now.
267 Still collecting temp files that nobody cares to depollute.
269 2009-03-21 Andreas J. Koenig <andk@cpan.org>
271 * Bug?: should it be harder than it is atm to set the timestamp to the
274 * bug with native integers:
278 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
282 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
285 Native integer broke when native math was turned off. FIXED
287 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
288 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
291 Yes, records out of order:
294 447004 epoch: 995885533
295 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
298 447008 epoch: 995890358
299 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
302 447012 epoch: 995892221
303 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
306 FIXED with sanity check and later with the integer fix.
308 * Bug: want the index files in a .recent directory
310 * Bug: lots of dot files are not deleted in time
312 * possible test case: can a delete change the timestamp? This would
313 probably break the order of events.
315 2009-03-20 Andreas J. Koenig <andk@cpan.org>
317 * 1233701831.34486 what's so special about this number/string? It
319 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
320 atm lives in Y,Q, and Z.
322 It is the first entry after id/--skip-locking which has timestamp
323 1234164228.11325 which represents a file that doesn't exist anymore.
327 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
328 _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
329 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
330 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
331 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
332 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
333 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
334 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
335 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
336 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
337 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
338 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
339 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
341 FIXED, it was the "--skip-locking" file where manual intervention was participating
343 * bug on the mirroring slave: when the dirtymark gets increased we
344 probably do not reset the done intervals. The mirrorer stays within
345 tight bounds where it tries to sync with upstream and never seems to
346 finish. In the debugging state file I see lots of identical intervals
347 that do not get collapsed. When I restart the mirrorer it dies with:
349 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
350 _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
351 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
352 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
353 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
354 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
355 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
356 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
357 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
358 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
360 and debugging stands at
366 epoch: 1237400802.5789
367 path: id/J/JO/JOHND/CHECKSUMS
370 epoch: 1237400807.97514
371 path: id/M/MI/MIYAGAWA/CHECKSUMS
374 epoch: 1237400817.94363
375 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
383 and it is reproducable.
385 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
386 pause has a fresh one.
388 FIXED, it was the third anded term in each of the ifs in the IV block in
389 _register_one: with that we make sure that we do not stamp on valuable
392 2009-03-17 Andreas J. Koenig <andk@cpan.org>
394 * done: verified the existence of the floating point bug in bleadperl
395 and verified that switching from YAML::Syck to YAML::XS does not resolve
398 BTW, the switch was doable with
400 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
402 and should be considered as a separate TODO
404 * todo: integrate a dirty update with two aggregate calls before
405 unlocking for frictionless dirtying DONE
407 * todo: start the second rsync daemon on pause DONE
409 * todo: move index files to .recent: this cannot simply be done by
410 setting filenameroot to .recent/RECENT. Other parts of the modules rely
411 on the fact that dirname(recentfile) is the root of the mirrored tree.
413 2009-03-16 Andreas J. Koenig <andk@cpan.org>
415 * What was the resolution of the mirror.pl delete hook bug? Do we call
416 the delete hook when pause removes a file from MUIR?
418 * Today on pause: Updating 2a13fba..29f284d and installing it for
419 /usr/local/perl-5.10.0{,-RC2}
421 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
422 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
424 2009-03-15 Andreas J. Koenig <andk@cpan.org>
426 * currently recent_events has the side effect of setting dirtymark
427 because it forces a read on the file. That should be transparent, so
428 that the dirtymark call always forces a cache-able(?) read.
430 * The bug below is -- after a lot of trying -- not reproducible on a
431 small script, only in the large test script. The closest to the output
437 my $x = "01237123229.8814";
442 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
444 $l = "1237123231.22458";
445 $r = "1237123231.22458";
447 Devel::Peek::Dump $l;
448 Devel::Peek::Dump $r;
449 Devel::Peek::Dump $x = $l <=> $r;
453 The checked in state at c404a85 fails the test with my
454 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
455 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
456 in the test script always consistent.
458 * Todo: write a test that inserts a second dirty file with an already
459 existing timestamp. DONE
461 * Bug in perl 5.10 on my 64bit box:
463 DB<98> Devel::Peek::Dump $l
464 SV = PVMG(0x19e0450) at 0x142a550
466 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
468 NV = 1237123231.22458
469 PV = 0x194ce70 "1237123231.22458"\0
473 DB<99> Devel::Peek::Dump $r
474 SV = PVMG(0x19e0240) at 0x142a3e8
476 FLAGS = (PADMY,POK,pPOK)
479 PV = 0x19ff900 "1237123231.22458"\0
483 DB<100> Devel::Peek::Dump $l <=> $r
484 SV = IV(0x19ea6e8) at 0x19ea6f0
486 FLAGS = (PADTMP,IOK,pIOK)
489 DB<101> Devel::Peek::Dump $l
490 SV = PVMG(0x19e0450) at 0x142a550
492 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
494 NV = 1237123231.22458
495 PV = 0x194ce70 "1237123231.22458"\0
499 DB<102> Devel::Peek::Dump $r
500 SV = PVMG(0x19e0240) at 0x142a3e8
502 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
504 NV = 1237123231.22458
505 PV = 0x19ff900 "1237123231.22458"\0
509 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
511 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
513 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
515 * $obj->merge ($other) needs to learn about equal epoch which may happen
516 since dirty_epoch intruded.
518 * Wontfix anytime soon: I think we currently do not support mkdir. Only
521 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
523 * Todo: continue working on update(...,$dirty_epoch). It must be
524 followed by a fast_aggregate! DONE
526 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
528 * maybe we need a closest_entry or fitting_interval or something like
529 that. We want to merge an event into the middle of some recentfile.
530 First we do not know which file, then we do not know where to lock,
531 where to enter the new item, when and where to correct the dirtymark.
533 So my thought is we should first find which file.
535 Another part of my brain answers: what would happen if we would enter
536 the new file into the smallest file just like an ordinary new event,
537 just as an old event?
539 (1) we would write a duplicate timestamp? No, this would be easy to
542 (2) we would make the file large quickly? Yes, but so what? We are
543 changing the dirtymark, so are willing to disturb the downstream hosts.
545 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
547 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
551 https://rt.cpan.org/Ticket/Display.html?id=41199
553 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
555 Strangest output being HAKANARDO who managed to upload
557 Here is a better oneliner that includes also the first line of each
560 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
562 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
564 * memory leak in the syncher? It currently weighs 100M.
568 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
571 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
573 * lookup by epoch and by path and use this ability on the pause to never
574 again register a file twice that doesn't need it. Let's call it
577 * after the dirtymark is done: fill up recentfiles with fake (historic)
578 entries; fill up with individual corrections; algorithm maybe to be done
579 with bigfloat so that we can always place something in the middle
580 between two entries. Before we must switch to bigfloat we could try to
581 use Data::Float::nextup to get the.
583 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
586 * dirtymark now lives in Recentfile, needs to be used in rmirror.
588 * find out why the downloader died after a couple of hours without a net
589 connection. Write a test that survives the not-existence of the other
592 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
594 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
595 it makes less sense with tight coupling for secondary files. But asking
596 the server what the current epoch is might be cheaper on HTTP than on
597 rsync. (Needs to be evaluated)
599 * remove the 0.00 from the verbose overview in the Merged column in the
602 * write tests that expose the problems of the last few days: cascading
603 client/server roles, tight coupling for secondary RFs, deletes after
606 * Some day we might want to have policy options for the slave:
607 tight/loose/no coupling with upstream for secondary RFs. tight is what
608 we have now. loose would wait until a gap occurs that can be closed.
610 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
612 * revisit all $rfs->[$i+1] places if they now make sense still
614 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
616 * another bug is the fact that the mirror command deletes files before
617 it unhides the index file, thus confusing downstream slaves. We must not
618 delete before unhiding and must delete after unhiding. FIXED.
620 * new complication about the slave that is playing a server role.
621 Currently we mirror from newest to oldest with a hidden temporary file
622 as index. And when one file is finished, we unhide the index file.
623 Imagine the cascading server/slave is dead for a day. It then starts
624 mirroring again with the freshest thing and unhides the freshest index
625 file when it has worked through it. In that moment it exposes a time
626 hole. Because it now works on the second recentfile which is still
629 We currently do nothing special to converge after such a drop out. At
630 least not intentionally and robustly and thought through.
632 The algorithm we use to seed the next file needs quite a lot of more
633 robustness than it currently has. Something to do with looking at the
634 merged element of the next rf and when it has dropped off, we seed
635 immediately. And if it ramains dropped off, we seed again, of course.
637 Nope, looking from smaller to larger RFS we look at the merged element
638 of this RF and at the minmax/max element of the next RF. If that
639 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
642 And we need a public accessor seed and unseed or seeded. But not the mix
643 of public and private stuff that then is used behind the back.
645 And then the secondary* stuff must go.
647 And we must understand what the impact is on the DONE system. Can it go
648 unnoticed that there was a hole? And could the DONE system have decided
649 the hole is covered? This should be testable with three directories where
650 the middle stops working for a while. Done->merge is suspicious, we must
651 stop it from merging non-conflatable neighbors due to broken continuity.
655 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
657 * Slaven suggests to have the current epoch or the whole current
658 recentfile available from the HTTP server and take it away with
659 keepalive. This direction goes the granularity down to subseconds.
661 We might want to rewrite everything to factor out transport and allow
662 the whole thing to run via HTTP.
664 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
666 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
669 * smoker on k81 fetching from k75 to verify cascading works. See
670 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
672 * maybe the loop should wait for CHECKSUMS file after every upload. And
673 CPAN.pm needs to deal with timestamps in the future.
675 * do not forget the dirtymark!
677 Text: have a new flag on recentfiles with the meaning: if this
678 changes, you're required to run a full rsync over all the files. The
679 reason why we set it would probably be: some foul happened. we injected
680 files in arbitrary places or didn't inject them although they changed.
681 The content of the flag? Timestamp? The relation between the
682 recentfiles would have to be inheritance from the principal, because any
683 out of band changes would soon later propagate to the next recentfile.
685 By upping the flag often one can easily ruin the slaves.
687 last out of band change? dirtymark?
689 Anyway, this implies that we read a potentially existing recentfile
692 And it implies that we have an eventloop that keeps us busy in 2-3
693 cycles, one for current stuff (tight loop) and one for the recentfiles
694 (cascade when principal has changed), one for the old stuff after a
697 And it implies that the out-of-band change in any of the recentfiles
698 must have a lock on the principal file and there is the place to set the
701 * start a FAQ, especially quick start guide questions. Also to aid those
702 problematic areas where we have no good solution, like the "links"
705 * wish feedback when we are slow.
709 * Remove a few DEBUG statements.
711 * The multiple-rrr way of doing things needs a new option to rmirror,
712 like piecemeal or so. Not urgent because after the first pass through,
713 things run smoothely. It's only ugly during the first pass.
715 * I have the suspicion that the code is broken that decides if the
716 neighboring RF needs to be seeded. I fear when too much time has gone
717 between two calls (in our case more than one hour), it would not seed
718 the neighbor. Of course this will never be noticed, so we need a good
721 * local/localroot confusion: I currently pass both options but one must
724 * accounts for early birds on PAUSE rsync daemon.
726 * hardcoded 20 seconds
728 * who mirrors the index? DOING now.
730 * which CPAN mirrors offer rsync?
732 * visit all XXX, visit all _float places
734 * rename the pathdb stuff, it's too confusing. No idea how.
736 * rrr-inotify, backpan, rrr-register
738 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
740 * current bugs: the pathdb seems to get no reset, the seeding of the
741 secondaryttl stuff seems not to have an effect. Have helped myself with
742 a rand(10), need to fix this back. So not checked in. Does the rand
745 The rand thing helps. The secondaryttl stuff was in the wrong line,
748 The pathdb stuff was because I called either _pathdb or __pathdb on the
749 wrong object. FIXED now.
751 * It's not so beautiful if we never fetch the recentfiles that are not
752 the principal, even if this is correct behaviour. We really do not need
753 them after we have fetched the whole content.
755 OK, we want a switch for that: secondaryttl DONE
757 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
759 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
760 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
761 it lets 40 follow. FIXED
763 * See that the long running process really only updates the principal
764 file unless it has missed a timespan during which something happened. If
765 nothing happened, it must notice even when it misses the timespan. DONE
767 * we must throw away the pathdb when we have reached the end of Z. From
768 that moment we can have a very small pathdb because the only reason for
769 a pathdb is that we know to ignore old records in old files. We won't
770 need this pathdb again before the next full pass over the data is
771 necessary and then we will rebuild it as we go along. DONE
773 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
775 * I think, Done::register_one is doing wrong in that it does not
776 conflate neighboring pieces. The covered() method cannot do this because
777 it has no recent_events array at hand. But register_one has it and could
778 do it and for some reason misses to do it (sometimes).
780 This means that the three tests I just wrote can probably not survive
781 because they test with an already broken Done structure.
783 The art now is to detect how it happens, then to reproduce, then write a
786 So from the logfile this is what happens: we have a good interval with
787 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
788 top of it. Locally we now mirror F2 and open a new done interval for it.
789 Then we mirror F1 but this time with the timestamp T1b. And when we then
790 try to close the gap, we do not find T1 but instead something older. We
791 should gladly accept this older piece and this would fix this bug.
795 * bug to fix: when the 1h file changes while rmirror is running, we do
796 correctly sync the new files but never switch to the 6h file but rather
797 stay in a rather quick loop that fetches the 1h file again and again.
799 Is it possible that we initialize a new object? Or does
800 get_remote_recentfile_as_tempfile overwrite something in myself?
802 Want a new option: _runstatusfile => $file which frequently dumps the
803 state of all recentfiles to a file.
807 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
809 * Todo: now teach update to verify the timestamp is about to write
810 against the previous and use _increase_a_bit if it doesn't comply with
811 strict monotony. DONE
813 * The problem of rounding. So far perl's default precision was
814 sufficient. One day it won't be. FakeFloat has an easy job when it is
815 only reading and other machines have written correctly. But when we want
816 to write a floating point number that is a bit larger than the other
817 one, then we need our own idea of precision.
819 Slaven said: just append a "1". This might be going towards the end of
820 usability too quickly. I'd like something that actually uses the decimal
821 system. Well, appending a 1 also does this but...
823 E.g. we have 1.0. nextup on this architecture is starting with
824 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
825 taken the 1.0000000000000003 and the next user comes and the time tells
826 him 1.0 again. He has to beat my number without stepping over the
827 nextup. This is much less space than I had when I chose 1,2,3.
829 What is also irritating is that nextup is architecture dependent. The
830 128 bit guy must choose very long numbers to fit in between whereas the
831 other one with 16 bit uses larger steps. But then the algorithm is the
832 same for both, so that would be a nice thing.
834 I see two situation where we need this. One is when Time::HiRes returns
835 us a value that is <= the last entry in our recentfile. In this case
836 (let's call it the end-case) we must fill the region between that number
837 and the next higher native floating point number. The other is when we
838 inject an old file into an old recentfile (we would then also set a new
839 dirtymark). We find the integer value already taken and need a slightly
840 different one (let's call it the middle-case). The difference between
841 the two situations is that the next user will want to find something
842 higher than my number in the end-case and something lower than my number
845 So I suggest we give the function both a value and an upper bound and it
846 calculates us a primitive middle. The upper bound in the middle-case is
847 the next integer. The upper bound on the end-case is the nextup floating
848 point number. But the latter poses another problem: if we have occupied
849 the middle m between x and nextup(x), then the nextup(m) will probably
850 not be the same as nextup(x) because some rounding will take place
851 before the nextup is calculated and when the rounding reaches the
852 nextup(x), we will end up at nextup(nextup(x)).
854 So we really need to consider the nextup and the nextdown from there and
855 then the middle and that's the number we may approach asymptotically.
858 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
860 * consider deprecating the use of RECENT.recent as a symlink. It turns
861 out to need extra hoops with the rsync options and just isn't worth it.
862 Or maybe these extra hoops are needed anyway for the rest of the tree?
863 Nope, can't be the case because not all filesystems support symlinks.
865 But before doing the large step, I'll deprecate the call of
866 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
867 only introduced to resolve RECENT.recent and complicates the routine far
868 beyond what it deserves.
870 DONE. Won't deprecate RECENT.recent, just moved its handling to the
873 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
875 * I think it's a bug that the rsync_option links must be set to true in
876 order to support RECENT.recent and that nobody cares to set it
877 automatically. Similar for ignore_link_stat_errors.
879 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
881 * Todo: find all todos together and make a plan what is missing for a
884 - verifytree or something like that. fsck maybe.
886 - rersyncrecent, the script itself? What it do?
888 - a way to only mirror the recentfiles without mirroring the whole
889 remote system such that people can decide to mirror only partially see
890 also 2008-08-30. .shadow-xxx directory? this also needed for a
891 filesystem that is still incomplete and might need the mirrorfiles for
894 - long living objects that mirror again and again. Inject something
895 into ta, see how it goes over to tb.
897 - how do we continue filling up the DONE system when we use an object
898 for the second time? "fully covered" and "uptodate" or new terminology.
900 - overview called on the wrong file should be understandable
902 - the meta data field that must change when we fake something up so that
903 the downstream people know they have to re-fetch everything.
905 - how tolerant are we against missing files upstream? how do we keep
906 track? there are legitimate cases where we did read upstream index right
907 before a file got deleted there and then find that file as new and want
908 it. There are other cases that are not self healing and must be tracked
911 - how, exactly, do we have to deal with deletes? With rsync errors?
913 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
914 authors) failed: No such file or directory (2)
916 The file above is a delete in 1h and a new in file 1M and the
917 delete in the locally running rmirror did not get propagated to the 1M
918 object. Bug. And the consequence is a standstill.
920 It seems that a slave that works with a file below the principal needs
921 to merge things all the way up to get rid of later deletes. Or keep
922 track of all deletes and skip them later. So we need a trackdeletes.pm
923 similar to the done.pm?
925 see also 2008-08-20 about spurious deletes that really have no add
926 counterpart and yet they are not wrong.
928 - consider the effect when resyncing the recentfile takes longer than
929 the time per loop. Then we never rsync any file. We need to diagnose
930 that and force an increase of that loop time. But when we later are fast
931 enough again because the net has recovered, then we need to switch back
932 to original parameters. ERm, no, it's enough to keep syncing at least
933 one file before refetching an index file.
935 - remember to verify that no temp files are left lying around and the
938 - status file for not long running jobs that want to track upstream with
941 - revisit all XXX _float areas and study Sub::Exporter DONE
943 - persistent DB even though we just said we do not need it. Just for
944 extended capabilities and time savings when, for example, upstream
945 announces a reset and we get new recentfiles and could then limit
946 ourselves to a subset of files (those that have a changed epoch) in a
947 first pass and would only then do the loop to verify the rest. Or
950 * Todo: aggregate files should know their feed and finding the principal
951 should be done stepwise. (?)
953 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
954 left locks around. DONE
956 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
958 * maybe extend the _overview so that it always says if and where the
959 last file is in the next file and where the next event in the next rf
960 would lie. No, don't like this anymore. REJECT
962 * take the two new redundant tests out again, only the third must
965 * Todo: add a sanity check if the merged structure is really pointing to
966 a different rf and that this different rf is larger. DONE
968 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
970 * now test, if they are overlapping. And test if there is a file in the
971 next rf that would fit into this rf's interval.
973 1h 1222324012.8474 1222322541.7963 0.4086
974 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
975 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
976 1W 1222313218.3626 1221708477.5829 167.9835
978 I suspect that somebody writes a merged timestamp without having merged
979 and then somebody else relies on it.
981 If aggregate is running, the intervals must not be extravagated, if it
982 is not running, there must not be bounds, the total number of events in
983 the system must be counted and must be controlled throughout the tests.
984 That the test required the additional update was probably nonsense,
985 because aggregate can cut pieces too. FIXED & DONE
987 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
989 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
992 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
994 * Most apparent bug at the moment is that the recentfiles are fetched
995 too often. Only the principal should be fetched and if it has not
996 changed, the others should not be refetched. ATM I must admit that I'm
997 happy that we refetch more often than needed because I can more easily
998 fix bugs while the thing is running.
1000 * Let's say, 1220474966.19501 is a timestamp of a file that is already
1001 done but the done system does not know about it. The reason for the
1002 failure is not known and we never reach the status uptodate because of
1003 this. We must get over it.
1005 Later it turns out that the origin server had a bug somewhere.
1006 1220474966.19042 came after 1220474966.19501. Or better: it was in the
1007 array of the recentfile one position above. The bug was my own.
1009 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1011 * There is the race condition where the server does a delete and the
1012 slave does not yet know and then tries to download it because he sees
1013 the new. So for this time window we must be more tolerant against
1014 failure. If we cannot download a file, we should just skip it and should
1015 not retry immediately. The whole system should discover the lost thing
1016 later. Keeping track with the DONE system should really be a no brainer.
1018 But there is something more: the whole filesystem is a database and the
1019 recentfiles are one possible representation of it. It's a pretty useful
1020 representation I think that's why I have implemented something around
1021 it. But for strictly local operation it has little value. For local
1022 operation we would much rather have a database. So we would enter every
1023 recentfile reading and every rsync operation and for every file the last
1024 state change and what it leads to. Then we would always ignore older
1025 records without the efforts involved with recentfiles.
1027 The database would have: path,recentepoch,rsyncedon,deletedon
1029 Oh well, not yet clear where this leads to.
1031 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1033 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
1034 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
1036 * statusfile, maybe only on demand, alone to have a sharp debugging
1037 tool. It is locked and all recentfiles dump themselves into it and we
1038 can build a viewer that lets us know where we stand and what's inside.
1040 * remember: only the principal recentfile needs expiration, all others
1041 shall be expired by principal if it discovers that something has move
1044 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1046 * Always check if we stringify to a higher value than in the entry
1049 * And in covered make an additional check if we would be able to see a
1050 numerical difference between the two numbers and if we can't then switch
1051 to a different, more expensive algorithm. Do not want to be caught by
1052 floating surprises. DONE
1054 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1056 * caching has several aspects here: we can cache the interval of the
1057 recentfile which only will change when the mtime of the file changes. We
1058 must re-mirror the recentfile when its ttl has expired. Does have_read
1059 tell you anything? It counts nothing at all. Only the mtime is
1060 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
1061 thing the have_mirrored because it is unrelated to the mtime.
1063 * Robustness of floating point calculations! I always thought that the
1064 string calculated by the origin server for the floating representation
1065 of the epoch time is just a string. When we convert it to a number and
1066 later back to a string, the other computer might come to a different
1067 conclusion. This must not happen, we want to preserve it under any
1068 circumstances. I will have to write tests with overlong sequences that
1069 get lost in arithmetic and must see if all still works well. DONE
1071 But one fragile point remains: if one host considers a>b and the other
1072 one considers them == but no eq. To prevent this, we must probably do
1073 some extra homework. DONE
1075 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1077 * the concept of tracking DONE needs an object per recentfile that has
1078 something like these methods:
1080 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1082 covered() register() covered()
1084 The unclear thing is how we translate points in time into intervals. We
1085 could pass a reference to the current recent_events array when running
1086 we_have(xxx) and let the DONE object iterate over it such that it only
1087 has to store a list of intervals that can melt into each other. Ah, even
1088 passing the list together with a list of indexes seems feasiable.
1090 Or maybe ask for the inverted list?
1092 Whenever the complete array is covered by the interval we say we are
1093 fully covered and if the recentfile is not expired, we are uptodate.
1095 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1097 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1099 * need a way to "return" the next entry after the end of a list. When
1100 the caller says "before" or "after" we would like to know if he could
1101 cover that interval/threshold or not because this influences the effect
1102 of a newer timestamp of that recentfile. DONE with $opt{info}.
1104 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1106 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1108 loop implies tixu (?). Nope, something like --statefile decides. Per
1111 T test, I init, X including removals, U nodirtymark
1113 So we have no concept of dirtymarks, we only trust that since we are
1114 running we have observed everything steadily. But people will not let
1115 this program run forever so we must consider both startup penalty and
1116 book keeping for later runs. We keep this for later. For now we write a
1117 long running mirror that merges several intervals.
1119 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1121 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1124 * rersyncrecent, the script: default to one week. The name of the switch
1125 is --after. Other switches? --loop!
1127 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1129 * need a switch --skip-deletes (?)
1131 * need a switch --enduser that tells us that the whole tempfile
1132 discipline is not needed when there is no downstream user. (?)
1134 Without this switch we cannot have a reasonable recent.pl that just
1135 displays the recent additions. Either we accept to download everything.
1136 Or we download temporary files without the typical rsync protocol
1139 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1140 File::Temp::tempdir, this might be a win.
1142 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1144 * apropos missing: we have no push, we never know the downstream
1145 servers. People who know their downstream hosts and want to ascertain
1146 something will want additional methods we have never thought about, like
1147 update or delete a certain file.
1149 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1151 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1152 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1155 * rsync options: it's a bit of a pain that we usually need several rsync
1156 options, like compress, links, times, checksum and that there is no
1157 reasonable default except the original rsync default. I think wee can
1158 safely assume that the rsync options are shared between all recentfile
1159 instances within one recent tree.
1161 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1163 * deletes: if a delete follows an add quickly enough it may happen that
1164 a downstream mirror did not see the add at all! It seems this needs to
1165 be mentioned somewhere. The point here is that even if the downstream is
1166 never missing the principal timeframe it may encounter a "delete" that
1167 has no complimentary "add" anywhere.
1169 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1171 * I suspect the treat of metadata is incorrect during read or something.
1172 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1173 contained more than 6 hours worth of data. At 06:08 we merged into the
1174 1d file. We need to take snapshots of the 6h file over the course of an
1175 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1178 Much worse: watching the 1h file: right at the moment (at 06:35) it
1179 covers 1218867584-1219120397 which is 70 hours.
1181 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1182 08:19:44 2008, that is when I checked out last time, so it seems to be
1183 aggregating and never truncating?
1185 No, correct is: it is never truncating; but wrong is: it is aggregating.
1186 It does receive a lot of events from time to time from a larger file.
1187 Somehow a large file gets merged into the small one and because the
1188 "meta/merged" attribute is missing, nobody is paying attention. I
1189 believe that I can fix this by making sure that metadata are honoured
1190 during read. DONE and test adjusted.
1192 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1194 * grand renaming plan
1196 remotebase => remoteroot to fit well with localroot DONE
1197 local_path() => localroot seems to me should already work DONE
1198 recentfile_basename => rfilename no need to stress it has no slash DONE
1200 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1201 kern? It would anyway need a deprecation cycle because it is an
1202 important constructor.
1204 * I like the portability that Data::Serializer brings us but the price
1205 is that some day we might find out that it is slowing us a bit. We'll
1208 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1210 * should we not enter the interval of the principal (or the interval of
1211 the merging file?) in every aggregated/merged file?
1213 * we should aim at a first release and give up on thinking about
1214 sanitizing stuff and zloop. Let's just admit that a full traditional
1215 rsync is the only available sanitizer ATM. Otherwise it's complicated
1216 stuff: sanitizing on the origin server, sanitizing on the slaves,
1217 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1218 get the basics out before this becomes a major cause for mess.
1220 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1222 * On OSes not supporting symlinks we expect that RECENT.recent contains
1223 the contents of the principal recentfile. Actually this is identical on
1224 systems supporting symlinks. Simple, what follows from that is that we
1225 need to keep the serializer in the metadata because we cannot read it
1226 from the filename, doesn't it? Of course not. It's a chicken and egg
1227 problem. This leaves us with the problem to actually parse the
1228 serialized data to find out in which format it is. So who can do the 4
1229 or 5 magics we wanted to support? File::LibMagic?
1231 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1233 * remotebase and recentfile_basename are ugly names. Now that we need a
1234 word for the shortest/principal/driving recentfile too we should do
1237 localroot is good. rfile is good. local_path() is bad, local_path($path)
1238 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1241 Up to now remotebase was the string that described the remote root
1242 directory in rsync notation, like pause.perl.org::authors. And
1243 recentfile_basename was "RECENT-1h.yaml".
1245 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1247 * The test that was added in today's checkin is a good start for a test
1248 of rmirror. We should have more methods in Recent.pm: verify,
1249 addmissingfiles. We should verify the current tree, then rmirror it and
1250 then verifytree the copy. We could then add some arbitrary file and let
1251 it be discovered by addmissingfiles, then rmirror again and then
1252 verifytree the copy again.
1254 Then we could start stealing from csync2 sqlite database [no port to
1255 OSX!] and fill a local DB. And methods to compare the database with the
1256 recentfiles. Our strength is that in principle we could maintain state
1257 with a single float. We have synced up to 1234567890.123456. If the Z
1258 file does not add new files all we have to do is mirror the new ones and
1261 This makes it clear that we should extend current protocol and declare
1262 that we cheat when we add files too late, just to help the other end
1263 keeping track. Ah yes, that's what was meant when zloop was mentioned
1266 Maybe need to revisit File::Mirror to help me with this task.
1268 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1270 * There must be an allow-me-to-truncate flag in every recentfile.
1271 Without it one could construct a sequence of updates winning the locking
1272 battle against the aggregator. Only if an aggregator has managed to
1273 merge data over to the next level, truncating can be allowed. DONE with
1276 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1278 * We should probably guarantee that no duplicates enter the aggregator
1281 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1283 * To get merge operation faster would need a good benchmark test. What
1284 02 spits out isn't reliable enough and is dominated by many other
1287 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1288 Author: Andreas J. Koenig <k@k75.(none)>
1289 Date: Sat Aug 2 07:58:04 2008 +0200
1293 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1294 Author: Andreas J. Koenig <k@k75.(none)>
1295 Date: Sat Aug 2 11:40:29 2008 +0200
1297 there was a lot of trying but the effect is hardly measurable with
1300 * overhead of connecting seems high. When setting
1301 max_files_per_connection to 1 we see that.
1303 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1305 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1307 25138.2215600014/3600 = 6.98283932222261
1309 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1310 be a bug in the aggregator. FIXED
1312 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1314 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1315 we have it, can we mirror it, mirror it!
1317 I fear this needs a new class which might be called
1318 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1319 and treat them as an entity. I realize that a single recentfile may be
1320 sufficient for certain tasks and that it is handy for the low level
1321 programmer but it is not nice to use. If there is a delete in the 1h
1322 file then the 6h file still contains it. Seekers of the best information
1323 need to combine at least some of the recentfiles most of the time.
1325 There is the place for the Z loop!
1327 But the combination is something to collect in a database, isn't it. Did
1328 csync2 just harrumph?
1330 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1332 * it just occurred to me that hosts in the same mirroring pool could
1333 help out each other even without rewriting the recentfile. Just fetch
1334 the stuff to mirror from several places, bingo. But that's something
1335 that should rather live in a separate package or in rsync directly.
1337 * cronjobs are unsuited because with ntp they would all come at the full
1338 minute and disturb each other. Besides that I'd hate to have a backbone
1339 with more than a few seconds latency.
1341 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1343 * a second rsync server with access control for PAUSE. Port? 873 is the
1344 standard port, let's take 8873.
1346 * if there were a filesystem based on this, it would have a slow access
1347 to inexistent files. It would probably provide wrong readdir (only based
1348 on current content) or also a slow one (based on a recentfile written
1349 after the call). But it would provide fast access to existing files. Or
1350 one would deliberately allow slightly blurred answers based on some
1351 sqlite reflection of the recentfiles.
1353 * todo: write a variant of mirror() that combines two or more
1354 recentfiles and treats them like one
1356 * todo: signal handler to remove the tempfile
1358 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1360 * now that we have the symlink I forgot how it should be used in
1363 * the z loop: add missing files to Z file. Just append them (instead of
1364 prepending). So one guy prepends something from the Y file from time to
1365 time and another guy appends something rather frequently. Collecting
1366 pond. When Y merges into Z, things get epoch and the collecting pond
1367 gets smaller. What exactly are "missing files"?
1369 take note of current epoch of the alpha file, let's call it the
1372 find all files on disk
1374 remove all files registered in the recentworld up to recent-ts
1376 remove all files that have been deleted after recent-ts according to
1379 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1381 * rersyncrecent might be a cronjob with a (locked) state file which
1382 contains things like after and maybe last z sync or such?
1384 rrr-mirror might be an alternative name but how would we justify the
1385 three Rs when there is no Re-Rsync-Recent?
1387 With the --loop parameter it is an endless loop, without it is no loop.
1388 At least this is simple.
1390 * todo: new accssor z-interval specifies how often the Z file is updated
1391 against the filesystem. We probably want no epoch stamp on these
1392 entries. And we want to be able to filter the entries (e.g. no
1393 by-modules and by-category tree)
1395 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1397 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1398 into Z. We do not need the other files filled up with filesystem
1401 * need interface to query for a file in order to NOT call update on
1402 PAUSE a second time within a short time.
1404 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1406 * recommended update interval? Makes no sense, is different for
1413 change-log-default-name: "Todo"