1 2011-02-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * IN_Q_OVERFLOW needs to trigger an fsck, likewise the entry into the
6 * wishlist: callback in recent_events that fires when an event is added
7 to the list so we can display the progress during fsck.
9 * speed: a large directory remove that is done by the kernel in 1
10 seconds triggers minutes of bookkeeping (07:56:42 - 08:01:10 for ~3000
11 files, all due to Schlemiehl again. Want to do some collecting of
12 pending ops before actually writing to the RECENT files. Lock!
14 2011-01-30 Andreas J. Koenig <andk@cpan.org>
16 * rrr-init and rrr-server need to be used such:
22 Stupid but no showstopper ATM.
24 2010-10-26 Andreas J. Koenig <andk@cpan.org>
26 * forking slave is done, needs more testing.
28 2010-07-06 Andreas J. Koenig <andk@cpan.org>
30 * The two next most important goals are to get a useable rrr-server and
31 a forking slave implementation that uses virtually no memory after the
32 first pass through. Does it need the status file?
34 * rrr-init now being fast we can now remove the alpha alerts. Done.
36 2010-06-16 Andreas J. Koenig <andk@cpan.org>
38 * rrr-init seems to be sssslllloooowwww. At 3:49 I started it for the
39 video tree. After twelve minutes the file has reached a size of 500k. It
40 is locking and rewriting the YAML after every file. Wasn't there an
41 unlocked variant? For "init" one would not expect locked correctness and
42 a rewrite after every item. Given that this tends to become slower with
43 every file it is really unacceptable.
45 Reminds me of rrr-dirtyupdate (the former name of rrr-update). It seems
46 the code was not dirty enough and that was the reason why it was so
47 slow. And rrr-update does not have a --dirty switch, so it looks like we
48 have no way to force dirty and fast operation yet but we need it for
51 Irritating the bootstrapping race: the order {(1) fill initial
52 array, (2) start server} leaves the need for an fsck to cover the time
53 between 1 and 2. Given that regular fsck is needed in any case the price
54 is not high but still doesn't give a warm feeling. Maybe starting the
55 server will by itself do an fsck?
57 2010-06-15 Andreas J. Koenig <andk@cpan.org>
59 * remove the wild alpha alerts and say, it is alpha but parts of it are
60 already working very well or so.
62 2009-12-10 Andreas J. Koenig <andk@cpan.org>
64 * Prominent request for statusfile to safe memory consumption. I'd like
65 to work backwards: get a verification program that verifies how well in
66 sync two nodes are without using frmr itself. Pretend that the forking
67 agent already works and can be interrupted with ^C. Then let it die at
68 random and restart at random and see the memory consumption and the
69 failure rate over time. Compare it with the curve we have now.
71 Of course we also invented the status file as a sharp debugging tool.
72 I'd like to see it in action. See also: F:R:M:Recent::thaw().
74 2009-06-22 Andreas J. Koenig <andk@cpan.org>
76 * Todo: rrr-init to generate some index files so that rrr-server can be
79 2009-05-02 Andreas J. Koenig <andk@cpan.org>
81 * DONE: rename rrr-dirtyupdate to rrr-update and add an option --dirty
82 so we can use it for normal update with a single file. And add a cronjob
83 that runs rrr-fsck so we get a reminder when something is not working
86 2009-04-27 Andreas J. Koenig <andk@cpan.org>
88 * Todo: I still see more link_stat errors than expected. I come back to
89 the former suspicion that we do not do necessary bookkeeping on deleted
90 files so that when we reach the "new" event on a meanwhile deleted file
91 we try to mirror it although we could know better. It's just noise, no
92 harm but noise is irritating.
94 * Todo: more experimenting with runstatusfile to get it official
96 * Todo: make sure $ENV{LANG}="C" when we call rsync because we will
99 rsync: link_stat "/authors/id/T/TH/THINC/DateTime-Format-Flexible-0.02.tar.gz" (in PAUSE) failed: No such file or directory
101 rsync: link_stat "/authors/id/A/AG/AGENT/SSH-Batch-0.001.meta" (in PAUSE) failed: No such file or directory (2)
103 And that ignore_link_stat_error is propagated in sparse_clone.
107 For the record: the bug was that ignore_link_stat_error was lost during
108 sparse_clone. And then the second bug was that it did not default to
109 true. Both are now fixed.
111 2009-04-25 Andreas J. Koenig <andk@cpan.org>
113 * Todo: fill rrr program with life
115 2009-04-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
117 * Minor bug: I see complaints about files not existing like
118 /authors/id/G/GW/GWADEJ/SVG-Sparkline-0.2.0.meta or
119 /authors/id/R/RC/RCAPUTO/POE-1.004.readme
121 In the RECENT files these are delete events but of course also new
122 events in older recent files.
124 I think we do not properly remember deletes when we skip-deletes and so
125 do not filter them out when we later see the "new" event.
127 Yes, but that what ignore_link_stat_error is for, we know about this
128 race condition. We must make ignore_link_stat_error default to true.
132 * bug in Done.pm failing to merge two fields into one when a third is
133 present. FIXED and accompanied with new test.
135 2009-04-24 Andreas J. Koenig <andk@cpan.org>
137 * Todo: keep the index clean and run some kind of nondestructive fsck 4
140 * who is our backbone?
142 http://cpan.cpantesters.org/authors/02STAMP # barbie
143 http://cpan.solfo.com/authors/02STAMP # abh
144 (http|ftp)://theoryx5.uwinnipeg.ca/pub/CPAN/ # rkobes
146 2009-04-17 Andreas J. Koenig <andk@cpan.org>
148 * the changelog helper I included in the Makefile release memo does not
149 show tags and is much less useful than I thought. gitk is probably much
152 2009-04-16 Andreas J. Koenig <andk@cpan.org>
156 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
157 #### Ydr_.yaml ... DONE
158 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
159 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
160 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
161 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
162 #### gone already at cpan-pause.pl line 0
164 Running without skip-deletes now but cannot reproduce.
166 2009-04-15 Andreas J. Koenig <andk@cpan.org>
168 * consider state directory such we can restart after a ^C or come again
171 Reuse _runstatusfile: write atomically, see how we can use the dump to
172 start running again. Do not leak and remove globs before writing. Do not
173 only dump one rf, dump the whole complex.
175 * consider the effect of ^C. Is it safe?
177 * want to use Perl::Version to up the versions on every release. Only if
178 things have changed, of course. Does this actually work?
180 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
181 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
183 Must do the bookkeeping myself when this should happen, maybe based on git?
185 I think I prefer the same version number for all pm files. But one also
186 needs a routine for setting them. Ahh:
188 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
194 cpan[2]> m /Mirror::Recent/
195 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
196 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
197 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
198 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
201 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
203 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
205 I just tried MM 6.5102 and it again produces only a provides for Recent,
206 not for the other modules.
208 My own fault. FIXED in the Makefile.PL.
210 2009-04-13 Andreas J. Koenig <andk@cpan.org>
212 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
214 * is it true (as stated at
215 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
216 that rsync can lead to errors when upstream changes before client sync
217 has completed? Can we see the error when we run this on upstream:
220 use Time::HiRes qw(time);
222 open my $fh, ">", "changingfile.txt.new" or die;
223 print $fh (time."\n") x 1000000;
225 rename "changingfile.txt.new", "changingfile.txt" or die;
229 And on the receiving end:
232 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
236 I cannot get it to fail. Apparently it is sufficient when upstream
237 always writes atomically (which of course is mandatory)?
239 SENT to mailing list Apr 28.
243 > https://fedorahosted.org/InstantMirror/
244 > https://www.redhat.com/mailman/listinfo/instantmirror-list
245 > irc.freenode.net channel #instantmirror
247 A cool name for a project. Inspires me to write a few sentences about
248 bittorrent's role in the grand picture.
250 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
251 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
253 2009-04-12 Andreas J. Koenig <andk@cpan.org>
255 * Interesting last minute bug during real download testing: the output
256 isn't as pretty anymore, it seems that more work is being done than
259 Introducing a highlevel _alluptodate method helped a lot in debugging
262 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
263 different on 1h file than on the other files. This lead to frequent
264 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
267 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
269 * Todo: try if we can be faster with a native float package. I'm really
270 glad to have a machine exposing perl floating point bugs, it forced me
271 to retract this item from the changes file: "several speedups in the
272 fakebigfloat code" before I get cpan testers fails.
274 2009-04-11 Andreas J. Koenig <andk@cpan.org>
276 * Bug: when the dirtymark changes upstream, then the downstream server
277 notices it immediately and mirrors the "1h" file but then leaves the
278 rmirror loop and continues with "6h" on the next round through. This
279 behaviour goes on until it reaches the Z file. Because we eat the Z file
280 piecemeal we leave the loop after a while but after that everybody seems
281 to have forgotten that there is some work left to be done. Bad, bad,
284 ETOOCONFUSING better name?
286 Recentfile::get_remote_recentfile_as_tempfile OK
287 Recentfile::resolve_recentfilename split_rfilename DONE
289 Recent::_principal_recentfile_object _principal_recentfile_fromremote
290 Recent::principal_recentfile OK
291 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
293 some void, some not, some not void but called in void context.
295 2009-04-10 Andreas J. Koenig <andk@cpan.org>
297 * Bug? When the dirtymark changes then the second tier hosts quickly
298 reset their DONE state and restart mirroring. But then they mirror
299 potentially outdated index files with inconsistent dirtymark because the
300 first tier box may be a bit behind on the large recentfiles.
302 This means we need a brake when iterating over the recentfiles that
303 refuses to use a recentfile with the wrong dirtymark.
305 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
307 * install the bin/* files and talk about rrr-overview somewhere.
309 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
310 created outside the target tree.
312 * Barbie asks for better logging capabilities
314 2009-03-29 Andreas J. Koenig <andk@cpan.org>
316 * known bug: cannot configure the Recentfile objects to keep delete
317 events. How to call the accessor for that? keep_delete_objects_forever?
319 Sounds acceptable. FIXED (but untested).
321 * known bug: keeps temporary index files lying around.
323 2009-03-24 Andreas J. Koenig <andk@cpan.org>
325 * Some equivalent for
327 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
331 2009-03-22 Andreas J. Koenig <andk@cpan.org>
333 * k81 is again client of k75 with different parameters.
335 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
336 about it all the time and nobody reacts accordingly; (2) lots of temp
337 files get created and not removed; culprit the new call to
338 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
339 caller has to remove the tempfile after use.
341 Need the drawing board.
343 PARTLY FIXED: retracting the idea to call
344 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
345 and unseed such that uptodate now means "we have mirrored this
346 recentfile's bunch of files" and seeded means "somebody believes the
347 index file for this rf needs be refreshed". This seems to work now.
349 Still collecting temp files that nobody cares to depollute.
351 2009-03-21 Andreas J. Koenig <andk@cpan.org>
353 * Bug?: should it be harder than it is atm to set the timestamp to the
356 * bug with native integers:
360 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
364 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
367 Native integer broke when native math was turned off. FIXED
369 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
370 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
373 Yes, records out of order:
376 447004 epoch: 995885533
377 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
380 447008 epoch: 995890358
381 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
384 447012 epoch: 995892221
385 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
388 FIXED with sanity check and later with the integer fix.
390 * Bug: want the index files in a .recent directory
392 * Bug: lots of dot files are not deleted in time
394 * possible test case: can a delete change the timestamp? This would
395 probably break the order of events.
397 2009-03-20 Andreas J. Koenig <andk@cpan.org>
399 * 1233701831.34486 what's so special about this number/string? It
401 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
402 atm lives in Y,Q, and Z.
404 It is the first entry after id/--skip-locking which has timestamp
405 1234164228.11325 which represents a file that doesn't exist anymore.
409 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
410 _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
411 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
412 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
413 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
414 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
415 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
416 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
417 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
418 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
419 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
420 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
421 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
423 FIXED, it was the "--skip-locking" file where manual intervention was participating
425 * bug on the mirroring slave: when the dirtymark gets increased we
426 probably do not reset the done intervals. The mirrorer stays within
427 tight bounds where it tries to sync with upstream and never seems to
428 finish. In the debugging state file I see lots of identical intervals
429 that do not get collapsed. When I restart the mirrorer it dies with:
431 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
432 _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
433 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
434 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
435 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
436 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
437 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
438 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
439 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
440 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
442 and debugging stands at
448 epoch: 1237400802.5789
449 path: id/J/JO/JOHND/CHECKSUMS
452 epoch: 1237400807.97514
453 path: id/M/MI/MIYAGAWA/CHECKSUMS
456 epoch: 1237400817.94363
457 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
465 and it is reproducable.
467 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
468 pause has a fresh one.
470 FIXED, it was the third anded term in each of the ifs in the IV block in
471 _register_one: with that we make sure that we do not stamp on valuable
474 2009-03-17 Andreas J. Koenig <andk@cpan.org>
476 * done: verified the existence of the floating point bug in bleadperl
477 and verified that switching from YAML::Syck to YAML::XS does not resolve
480 BTW, the switch was doable with
482 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
484 and should be considered as a separate TODO
486 * todo: integrate a dirty update with two aggregate calls before
487 unlocking for frictionless dirtying DONE
489 * todo: start the second rsync daemon on pause DONE
491 * todo: move index files to .recent: this cannot simply be done by
492 setting filenameroot to .recent/RECENT. Other parts of the modules rely
493 on the fact that dirname(recentfile) is the root of the mirrored tree.
495 2009-03-16 Andreas J. Koenig <andk@cpan.org>
497 * What was the resolution of the mirror.pl delete hook bug? Do we call
498 the delete hook when pause removes a file from MUIR?
500 * Today on pause: Updating 2a13fba..29f284d and installing it for
501 /usr/local/perl-5.10.0{,-RC2}
503 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
504 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
506 2009-03-15 Andreas J. Koenig <andk@cpan.org>
508 * currently recent_events has the side effect of setting dirtymark
509 because it forces a read on the file. That should be transparent, so
510 that the dirtymark call always forces a cache-able(?) read.
512 * The bug below is -- after a lot of trying -- not reproducible on a
513 small script, only in the large test script. The closest to the output
519 my $x = "01237123229.8814";
524 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
526 $l = "1237123231.22458";
527 $r = "1237123231.22458";
529 Devel::Peek::Dump $l;
530 Devel::Peek::Dump $r;
531 Devel::Peek::Dump $x = $l <=> $r;
535 The checked in state at c404a85 fails the test with my
536 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
537 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
538 in the test script always consistent.
540 * Todo: write a test that inserts a second dirty file with an already
541 existing timestamp. DONE
543 * Bug in perl 5.10 on my 64bit box:
545 DB<98> Devel::Peek::Dump $l
546 SV = PVMG(0x19e0450) at 0x142a550
548 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
550 NV = 1237123231.22458
551 PV = 0x194ce70 "1237123231.22458"\0
555 DB<99> Devel::Peek::Dump $r
556 SV = PVMG(0x19e0240) at 0x142a3e8
558 FLAGS = (PADMY,POK,pPOK)
561 PV = 0x19ff900 "1237123231.22458"\0
565 DB<100> Devel::Peek::Dump $l <=> $r
566 SV = IV(0x19ea6e8) at 0x19ea6f0
568 FLAGS = (PADTMP,IOK,pIOK)
571 DB<101> Devel::Peek::Dump $l
572 SV = PVMG(0x19e0450) at 0x142a550
574 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
576 NV = 1237123231.22458
577 PV = 0x194ce70 "1237123231.22458"\0
581 DB<102> Devel::Peek::Dump $r
582 SV = PVMG(0x19e0240) at 0x142a3e8
584 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
586 NV = 1237123231.22458
587 PV = 0x19ff900 "1237123231.22458"\0
591 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
593 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
595 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
597 * $obj->merge ($other) needs to learn about equal epoch which may happen
598 since dirty_epoch intruded.
600 * Wontfix anytime soon: I think we currently do not support mkdir. Only
603 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
605 * Todo: continue working on update(...,$dirty_epoch). It must be
606 followed by a fast_aggregate! DONE
608 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
610 * maybe we need a closest_entry or fitting_interval or something like
611 that. We want to merge an event into the middle of some recentfile.
612 First we do not know which file, then we do not know where to lock,
613 where to enter the new item, when and where to correct the dirtymark.
615 So my thought is we should first find which file.
617 Another part of my brain answers: what would happen if we would enter
618 the new file into the smallest file just like an ordinary new event,
619 just as an old event?
621 (1) we would write a duplicate timestamp? No, this would be easy to
624 (2) we would make the file large quickly? Yes, but so what? We are
625 changing the dirtymark, so are willing to disturb the downstream hosts.
627 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
629 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
633 https://rt.cpan.org/Ticket/Display.html?id=41199
635 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
637 Strangest output being HAKANARDO who managed to upload
639 Here is a better oneliner that includes also the first line of each
642 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
644 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
646 * memory leak in the syncher? It currently weighs 100M.
650 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
653 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
655 * lookup by epoch and by path and use this ability on the pause to never
656 again register a file twice that doesn't need it. Let's call it
659 * after the dirtymark is done: fill up recentfiles with fake (historic)
660 entries; fill up with individual corrections; algorithm maybe to be done
661 with bigfloat so that we can always place something in the middle
662 between two entries. Before we must switch to bigfloat we could try to
663 use Data::Float::nextup to get the.
665 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
668 * dirtymark now lives in Recentfile, needs to be used in rmirror.
670 * find out why the downloader died after a couple of hours without a net
671 connection. Write a test that survives the not-existence of the other
674 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
676 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
677 it makes less sense with tight coupling for secondary files. But asking
678 the server what the current epoch is might be cheaper on HTTP than on
679 rsync. (Needs to be evaluated)
681 * remove the 0.00 from the verbose overview in the Merged column in the
684 * write tests that expose the problems of the last few days: cascading
685 client/server roles, tight coupling for secondary RFs, deletes after
688 * Some day we might want to have policy options for the slave:
689 tight/loose/no coupling with upstream for secondary RFs. tight is what
690 we have now. loose would wait until a gap occurs that can be closed.
692 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
694 * revisit all $rfs->[$i+1] places if they now make sense still
696 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
698 * another bug is the fact that the mirror command deletes files before
699 it unhides the index file, thus confusing downstream slaves. We must not
700 delete before unhiding and must delete after unhiding. FIXED.
702 * new complication about the slave that is playing a server role.
703 Currently we mirror from newest to oldest with a hidden temporary file
704 as index. And when one file is finished, we unhide the index file.
705 Imagine the cascading server/slave is dead for a day. It then starts
706 mirroring again with the freshest thing and unhides the freshest index
707 file when it has worked through it. In that moment it exposes a time
708 hole. Because it now works on the second recentfile which is still
711 We currently do nothing special to converge after such a drop out. At
712 least not intentionally and robustly and thought through.
714 The algorithm we use to seed the next file needs quite a lot of more
715 robustness than it currently has. Something to do with looking at the
716 merged element of the next rf and when it has dropped off, we seed
717 immediately. And if it ramains dropped off, we seed again, of course.
719 Nope, looking from smaller to larger RFS we look at the merged element
720 of this RF and at the minmax/max element of the next RF. If that
721 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
724 And we need a public accessor seed and unseed or seeded. But not the mix
725 of public and private stuff that then is used behind the back.
727 And then the secondary* stuff must go.
729 And we must understand what the impact is on the DONE system. Can it go
730 unnoticed that there was a hole? And could the DONE system have decided
731 the hole is covered? This should be testable with three directories where
732 the middle stops working for a while. Done->merge is suspicious, we must
733 stop it from merging non-conflatable neighbors due to broken continuity.
737 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
739 * Slaven suggests to have the current epoch or the whole current
740 recentfile available from the HTTP server and take it away with
741 keepalive. This direction goes the granularity down to subseconds.
743 We might want to rewrite everything to factor out transport and allow
744 the whole thing to run via HTTP.
746 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
748 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
751 * smoker on k81 fetching from k75 to verify cascading works. See
752 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
754 * maybe the loop should wait for CHECKSUMS file after every upload. And
755 CPAN.pm needs to deal with timestamps in the future.
757 * do not forget the dirtymark!
759 Text: have a new flag on recentfiles with the meaning: if this
760 changes, you're required to run a full rsync over all the files. The
761 reason why we set it would probably be: some foul happened. we injected
762 files in arbitrary places or didn't inject them although they changed.
763 The content of the flag? Timestamp? The relation between the
764 recentfiles would have to be inheritance from the principal, because any
765 out of band changes would soon later propagate to the next recentfile.
767 By upping the flag often one can easily ruin the slaves.
769 last out of band change? dirtymark?
771 Anyway, this implies that we read a potentially existing recentfile
774 And it implies that we have an eventloop that keeps us busy in 2-3
775 cycles, one for current stuff (tight loop) and one for the recentfiles
776 (cascade when principal has changed), one for the old stuff after a
779 And it implies that the out-of-band change in any of the recentfiles
780 must have a lock on the principal file and there is the place to set the
783 * start a FAQ, especially quick start guide questions. Also to aid those
784 problematic areas where we have no good solution, like the "links"
787 * wish feedback when we are slow.
791 * Remove a few DEBUG statements.
793 * The multiple-rrr way of doing things needs a new option to rmirror,
794 like piecemeal or so. Not urgent because after the first pass through,
795 things run smoothely. It's only ugly during the first pass.
797 * I have the suspicion that the code is broken that decides if the
798 neighboring RF needs to be seeded. I fear when too much time has gone
799 between two calls (in our case more than one hour), it would not seed
800 the neighbor. Of course this will never be noticed, so we need a good
803 * local/localroot confusion: I currently pass both options but one must
806 * accounts for early birds on PAUSE rsync daemon.
808 * hardcoded 20 seconds
810 * who mirrors the index? DOING now.
812 * which CPAN mirrors offer rsync?
814 * visit all XXX, visit all _float places
816 * rename the pathdb stuff, it's too confusing. No idea how.
818 * rrr-inotify, backpan, rrr-register
820 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
822 * current bugs: the pathdb seems to get no reset, the seeding of the
823 secondaryttl stuff seems not to have an effect. Have helped myself with
824 a rand(10), need to fix this back. So not checked in. Does the rand
827 The rand thing helps. The secondaryttl stuff was in the wrong line,
830 The pathdb stuff was because I called either _pathdb or __pathdb on the
831 wrong object. FIXED now.
833 * It's not so beautiful if we never fetch the recentfiles that are not
834 the principal, even if this is correct behaviour. We really do not need
835 them after we have fetched the whole content.
837 OK, we want a switch for that: secondaryttl DONE
839 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
841 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
842 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
843 it lets 40 follow. FIXED
845 * See that the long running process really only updates the principal
846 file unless it has missed a timespan during which something happened. If
847 nothing happened, it must notice even when it misses the timespan. DONE
849 * we must throw away the pathdb when we have reached the end of Z. From
850 that moment we can have a very small pathdb because the only reason for
851 a pathdb is that we know to ignore old records in old files. We won't
852 need this pathdb again before the next full pass over the data is
853 necessary and then we will rebuild it as we go along. DONE
855 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
857 * I think, Done::register_one is doing wrong in that it does not
858 conflate neighboring pieces. The covered() method cannot do this because
859 it has no recent_events array at hand. But register_one has it and could
860 do it and for some reason misses to do it (sometimes).
862 This means that the three tests I just wrote can probably not survive
863 because they test with an already broken Done structure.
865 The art now is to detect how it happens, then to reproduce, then write a
868 So from the logfile this is what happens: we have a good interval with
869 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
870 top of it. Locally we now mirror F2 and open a new done interval for it.
871 Then we mirror F1 but this time with the timestamp T1b. And when we then
872 try to close the gap, we do not find T1 but instead something older. We
873 should gladly accept this older piece and this would fix this bug.
877 * bug to fix: when the 1h file changes while rmirror is running, we do
878 correctly sync the new files but never switch to the 6h file but rather
879 stay in a rather quick loop that fetches the 1h file again and again.
881 Is it possible that we initialize a new object? Or does
882 get_remote_recentfile_as_tempfile overwrite something in myself?
884 Want a new option: _runstatusfile => $file which frequently dumps the
885 state of all recentfiles to a file.
889 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
891 * Todo: now teach update to verify the timestamp is about to write
892 against the previous and use _increase_a_bit if it doesn't comply with
893 strict monotony. DONE
895 * The problem of rounding. So far perl's default precision was
896 sufficient. One day it won't be. FakeFloat has an easy job when it is
897 only reading and other machines have written correctly. But when we want
898 to write a floating point number that is a bit larger than the other
899 one, then we need our own idea of precision.
901 Slaven said: just append a "1". This might be going towards the end of
902 usability too quickly. I'd like something that actually uses the decimal
903 system. Well, appending a 1 also does this but...
905 E.g. we have 1.0. nextup on this architecture is starting with
906 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
907 taken the 1.0000000000000003 and the next user comes and the time tells
908 him 1.0 again. He has to beat my number without stepping over the
909 nextup. This is much less space than I had when I chose 1,2,3.
911 What is also irritating is that nextup is architecture dependent. The
912 128 bit guy must choose very long numbers to fit in between whereas the
913 other one with 16 bit uses larger steps. But then the algorithm is the
914 same for both, so that would be a nice thing.
916 I see two situation where we need this. One is when Time::HiRes returns
917 us a value that is <= the last entry in our recentfile. In this case
918 (let's call it the end-case) we must fill the region between that number
919 and the next higher native floating point number. The other is when we
920 inject an old file into an old recentfile (we would then also set a new
921 dirtymark). We find the integer value already taken and need a slightly
922 different one (let's call it the middle-case). The difference between
923 the two situations is that the next user will want to find something
924 higher than my number in the end-case and something lower than my number
927 So I suggest we give the function both a value and an upper bound and it
928 calculates us a primitive middle. The upper bound in the middle-case is
929 the next integer. The upper bound on the end-case is the nextup floating
930 point number. But the latter poses another problem: if we have occupied
931 the middle m between x and nextup(x), then the nextup(m) will probably
932 not be the same as nextup(x) because some rounding will take place
933 before the nextup is calculated and when the rounding reaches the
934 nextup(x), we will end up at nextup(nextup(x)).
936 So we really need to consider the nextup and the nextdown from there and
937 then the middle and that's the number we may approach asymptotically.
940 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
942 * consider deprecating the use of RECENT.recent as a symlink. It turns
943 out to need extra hoops with the rsync options and just isn't worth it.
944 Or maybe these extra hoops are needed anyway for the rest of the tree?
945 Nope, can't be the case because not all filesystems support symlinks.
947 But before doing the large step, I'll deprecate the call of
948 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
949 only introduced to resolve RECENT.recent and complicates the routine far
950 beyond what it deserves.
952 DONE. Won't deprecate RECENT.recent, just moved its handling to the
955 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
957 * I think it's a bug that the rsync_option links must be set to true in
958 order to support RECENT.recent and that nobody cares to set it
959 automatically. Similar for ignore_link_stat_errors.
961 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
963 * Todo: find all todos together and make a plan what is missing for a
966 - verifytree or something like that. fsck maybe.
968 - rersyncrecent, the script itself? What it do?
970 - a way to only mirror the recentfiles without mirroring the whole
971 remote system such that people can decide to mirror only partially see
972 also 2008-08-30. .shadow-xxx directory? this also needed for a
973 filesystem that is still incomplete and might need the mirrorfiles for
976 - long living objects that mirror again and again. Inject something
977 into ta, see how it goes over to tb.
979 - how do we continue filling up the DONE system when we use an object
980 for the second time? "fully covered" and "uptodate" or new terminology.
982 - overview called on the wrong file should be understandable
984 - the meta data field that must change when we fake something up so that
985 the downstream people know they have to re-fetch everything.
987 - how tolerant are we against missing files upstream? how do we keep
988 track? there are legitimate cases where we did read upstream index right
989 before a file got deleted there and then find that file as new and want
990 it. There are other cases that are not self healing and must be tracked
993 - how, exactly, do we have to deal with deletes? With rsync errors?
995 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
996 authors) failed: No such file or directory (2)
998 The file above is a delete in 1h and a new in file 1M and the
999 delete in the locally running rmirror did not get propagated to the 1M
1000 object. Bug. And the consequence is a standstill.
1002 It seems that a slave that works with a file below the principal needs
1003 to merge things all the way up to get rid of later deletes. Or keep
1004 track of all deletes and skip them later. So we need a trackdeletes.pm
1005 similar to the done.pm?
1007 see also 2008-08-20 about spurious deletes that really have no add
1008 counterpart and yet they are not wrong.
1010 - consider the effect when resyncing the recentfile takes longer than
1011 the time per loop. Then we never rsync any file. We need to diagnose
1012 that and force an increase of that loop time. But when we later are fast
1013 enough again because the net has recovered, then we need to switch back
1014 to original parameters. ERm, no, it's enough to keep syncing at least
1015 one file before refetching an index file.
1017 - remember to verify that no temp files are left lying around and the
1020 - status file for not long running jobs that want to track upstream with
1023 - revisit all XXX _float areas and study Sub::Exporter DONE
1025 - persistent DB even though we just said we do not need it. Just for
1026 extended capabilities and time savings when, for example, upstream
1027 announces a reset and we get new recentfiles and could then limit
1028 ourselves to a subset of files (those that have a changed epoch) in a
1029 first pass and would only then do the loop to verify the rest. Or
1032 * Todo: aggregate files should know their feed and finding the principal
1033 should be done stepwise. (?)
1035 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
1036 left locks around. DONE
1038 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1040 * maybe extend the _overview so that it always says if and where the
1041 last file is in the next file and where the next event in the next rf
1042 would lie. No, don't like this anymore. REJECT
1044 * take the two new redundant tests out again, only the third must
1047 * Todo: add a sanity check if the merged structure is really pointing to
1048 a different rf and that this different rf is larger. DONE
1050 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1052 * now test, if they are overlapping. And test if there is a file in the
1053 next rf that would fit into this rf's interval.
1055 1h 1222324012.8474 1222322541.7963 0.4086
1056 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
1057 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
1058 1W 1222313218.3626 1221708477.5829 167.9835
1060 I suspect that somebody writes a merged timestamp without having merged
1061 and then somebody else relies on it.
1063 If aggregate is running, the intervals must not be extravagated, if it
1064 is not running, there must not be bounds, the total number of events in
1065 the system must be counted and must be controlled throughout the tests.
1066 That the test required the additional update was probably nonsense,
1067 because aggregate can cut pieces too. FIXED & DONE
1069 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1071 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
1074 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1076 * Most apparent bug at the moment is that the recentfiles are fetched
1077 too often. Only the principal should be fetched and if it has not
1078 changed, the others should not be refetched. ATM I must admit that I'm
1079 happy that we refetch more often than needed because I can more easily
1080 fix bugs while the thing is running.
1082 * Let's say, 1220474966.19501 is a timestamp of a file that is already
1083 done but the done system does not know about it. The reason for the
1084 failure is not known and we never reach the status uptodate because of
1085 this. We must get over it.
1087 Later it turns out that the origin server had a bug somewhere.
1088 1220474966.19042 came after 1220474966.19501. Or better: it was in the
1089 array of the recentfile one position above. The bug was my own.
1091 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1093 * There is the race condition where the server does a delete and the
1094 slave does not yet know and then tries to download it because he sees
1095 the new. So for this time window we must be more tolerant against
1096 failure. If we cannot download a file, we should just skip it and should
1097 not retry immediately. The whole system should discover the lost thing
1098 later. Keeping track with the DONE system should really be a no brainer.
1100 But there is something more: the whole filesystem is a database and the
1101 recentfiles are one possible representation of it. It's a pretty useful
1102 representation I think that's why I have implemented something around
1103 it. But for strictly local operation it has little value. For local
1104 operation we would much rather have a database. So we would enter every
1105 recentfile reading and every rsync operation and for every file the last
1106 state change and what it leads to. Then we would always ignore older
1107 records without the efforts involved with recentfiles.
1109 The database would have: path,recentepoch,rsyncedon,deletedon
1111 Oh well, not yet clear where this leads to.
1113 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1115 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
1116 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
1118 * statusfile, maybe only on demand, alone to have a sharp debugging
1119 tool. It is locked and all recentfiles dump themselves into it and we
1120 can build a viewer that lets us know where we stand and what's inside.
1122 * remember: only the principal recentfile needs expiration, all others
1123 shall be expired by principal if it discovers that something has move
1126 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1128 * Always check if we stringify to a higher value than in the entry
1131 * And in covered make an additional check if we would be able to see a
1132 numerical difference between the two numbers and if we can't then switch
1133 to a different, more expensive algorithm. Do not want to be caught by
1134 floating surprises. DONE
1136 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1138 * caching has several aspects here: we can cache the interval of the
1139 recentfile which only will change when the mtime of the file changes. We
1140 must re-mirror the recentfile when its ttl has expired. Does have_read
1141 tell you anything? It counts nothing at all. Only the mtime is
1142 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
1143 thing the have_mirrored because it is unrelated to the mtime.
1145 * Robustness of floating point calculations! I always thought that the
1146 string calculated by the origin server for the floating representation
1147 of the epoch time is just a string. When we convert it to a number and
1148 later back to a string, the other computer might come to a different
1149 conclusion. This must not happen, we want to preserve it under any
1150 circumstances. I will have to write tests with overlong sequences that
1151 get lost in arithmetic and must see if all still works well. DONE
1153 But one fragile point remains: if one host considers a>b and the other
1154 one considers them == but no eq. To prevent this, we must probably do
1155 some extra homework. DONE
1157 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1159 * the concept of tracking DONE needs an object per recentfile that has
1160 something like these methods:
1162 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1164 covered() register() covered()
1166 The unclear thing is how we translate points in time into intervals. We
1167 could pass a reference to the current recent_events array when running
1168 we_have(xxx) and let the DONE object iterate over it such that it only
1169 has to store a list of intervals that can melt into each other. Ah, even
1170 passing the list together with a list of indexes seems feasiable.
1172 Or maybe ask for the inverted list?
1174 Whenever the complete array is covered by the interval we say we are
1175 fully covered and if the recentfile is not expired, we are uptodate.
1177 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1179 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1181 * need a way to "return" the next entry after the end of a list. When
1182 the caller says "before" or "after" we would like to know if he could
1183 cover that interval/threshold or not because this influences the effect
1184 of a newer timestamp of that recentfile. DONE with $opt{info}.
1186 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1188 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1190 loop implies tixu (?). Nope, something like --statefile decides. Per
1193 T test, I init, X including removals, U nodirtymark
1195 So we have no concept of dirtymarks, we only trust that since we are
1196 running we have observed everything steadily. But people will not let
1197 this program run forever so we must consider both startup penalty and
1198 book keeping for later runs. We keep this for later. For now we write a
1199 long running mirror that merges several intervals.
1201 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1203 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1206 * rersyncrecent, the script: default to one week. The name of the switch
1207 is --after. Other switches? --loop!
1209 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1211 * need a switch --skip-deletes (?)
1213 * need a switch --enduser that tells us that the whole tempfile
1214 discipline is not needed when there is no downstream user. (?)
1216 Without this switch we cannot have a reasonable recent.pl that just
1217 displays the recent additions. Either we accept to download everything.
1218 Or we download temporary files without the typical rsync protocol
1221 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1222 File::Temp::tempdir, this might be a win.
1224 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1226 * apropos missing: we have no push, we never know the downstream
1227 servers. People who know their downstream hosts and want to ascertain
1228 something will want additional methods we have never thought about, like
1229 update or delete a certain file.
1231 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1233 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1234 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1237 * rsync options: it's a bit of a pain that we usually need several rsync
1238 options, like compress, links, times, checksum and that there is no
1239 reasonable default except the original rsync default. I think wee can
1240 safely assume that the rsync options are shared between all recentfile
1241 instances within one recent tree.
1243 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1245 * deletes: if a delete follows an add quickly enough it may happen that
1246 a downstream mirror did not see the add at all! It seems this needs to
1247 be mentioned somewhere. The point here is that even if the downstream is
1248 never missing the principal timeframe it may encounter a "delete" that
1249 has no complimentary "add" anywhere.
1251 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1253 * I suspect the treat of metadata is incorrect during read or something.
1254 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1255 contained more than 6 hours worth of data. At 06:08 we merged into the
1256 1d file. We need to take snapshots of the 6h file over the course of an
1257 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1260 Much worse: watching the 1h file: right at the moment (at 06:35) it
1261 covers 1218867584-1219120397 which is 70 hours.
1263 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1264 08:19:44 2008, that is when I checked out last time, so it seems to be
1265 aggregating and never truncating?
1267 No, correct is: it is never truncating; but wrong is: it is aggregating.
1268 It does receive a lot of events from time to time from a larger file.
1269 Somehow a large file gets merged into the small one and because the
1270 "meta/merged" attribute is missing, nobody is paying attention. I
1271 believe that I can fix this by making sure that metadata are honoured
1272 during read. DONE and test adjusted.
1274 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1276 * grand renaming plan
1278 remotebase => remoteroot to fit well with localroot DONE
1279 local_path() => localroot seems to me should already work DONE
1280 recentfile_basename => rfilename no need to stress it has no slash DONE
1282 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1283 kern? It would anyway need a deprecation cycle because it is an
1284 important constructor.
1286 * I like the portability that Data::Serializer brings us but the price
1287 is that some day we might find out that it is slowing us a bit. We'll
1290 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1292 * should we not enter the interval of the principal (or the interval of
1293 the merging file?) in every aggregated/merged file?
1295 * we should aim at a first release and give up on thinking about
1296 sanitizing stuff and zloop. Let's just admit that a full traditional
1297 rsync is the only available sanitizer ATM. Otherwise it's complicated
1298 stuff: sanitizing on the origin server, sanitizing on the slaves,
1299 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1300 get the basics out before this becomes a major cause for mess.
1302 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1304 * On OSes not supporting symlinks we expect that RECENT.recent contains
1305 the contents of the principal recentfile. Actually this is identical on
1306 systems supporting symlinks. Simple, what follows from that is that we
1307 need to keep the serializer in the metadata because we cannot read it
1308 from the filename, doesn't it? Of course not. It's a chicken and egg
1309 problem. This leaves us with the problem to actually parse the
1310 serialized data to find out in which format it is. So who can do the 4
1311 or 5 magics we wanted to support? File::LibMagic?
1313 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1315 * remotebase and recentfile_basename are ugly names. Now that we need a
1316 word for the shortest/principal/driving recentfile too we should do
1319 localroot is good. rfile is good. local_path() is bad, local_path($path)
1320 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1323 Up to now remotebase was the string that described the remote root
1324 directory in rsync notation, like pause.perl.org::authors. And
1325 recentfile_basename was "RECENT-1h.yaml".
1327 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1329 * The test that was added in today's checkin is a good start for a test
1330 of rmirror. We should have more methods in Recent.pm: verify,
1331 addmissingfiles. We should verify the current tree, then rmirror it and
1332 then verifytree the copy. We could then add some arbitrary file and let
1333 it be discovered by addmissingfiles, then rmirror again and then
1334 verifytree the copy again.
1336 Then we could start stealing from csync2 sqlite database [no port to
1337 OSX!] and fill a local DB. And methods to compare the database with the
1338 recentfiles. Our strength is that in principle we could maintain state
1339 with a single float. We have synced up to 1234567890.123456. If the Z
1340 file does not add new files all we have to do is mirror the new ones and
1343 This makes it clear that we should extend current protocol and declare
1344 that we cheat when we add files too late, just to help the other end
1345 keeping track. Ah yes, that's what was meant when zloop was mentioned
1348 Maybe need to revisit File::Mirror to help me with this task.
1350 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1352 * There must be an allow-me-to-truncate flag in every recentfile.
1353 Without it one could construct a sequence of updates winning the locking
1354 battle against the aggregator. Only if an aggregator has managed to
1355 merge data over to the next level, truncating can be allowed. DONE with
1358 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1360 * We should probably guarantee that no duplicates enter the aggregator
1363 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1365 * To get merge operation faster would need a good benchmark test. What
1366 02 spits out isn't reliable enough and is dominated by many other
1369 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1370 Author: Andreas J. Koenig <k@k75.(none)>
1371 Date: Sat Aug 2 07:58:04 2008 +0200
1375 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1376 Author: Andreas J. Koenig <k@k75.(none)>
1377 Date: Sat Aug 2 11:40:29 2008 +0200
1379 there was a lot of trying but the effect is hardly measurable with
1382 * overhead of connecting seems high. When setting
1383 max_files_per_connection to 1 we see that.
1385 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1387 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1389 25138.2215600014/3600 = 6.98283932222261
1391 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1392 be a bug in the aggregator. FIXED
1394 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1396 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1397 we have it, can we mirror it, mirror it!
1399 I fear this needs a new class which might be called
1400 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1401 and treat them as an entity. I realize that a single recentfile may be
1402 sufficient for certain tasks and that it is handy for the low level
1403 programmer but it is not nice to use. If there is a delete in the 1h
1404 file then the 6h file still contains it. Seekers of the best information
1405 need to combine at least some of the recentfiles most of the time.
1407 There is the place for the Z loop!
1409 But the combination is something to collect in a database, isn't it. Did
1410 csync2 just harrumph?
1412 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1414 * it just occurred to me that hosts in the same mirroring pool could
1415 help out each other even without rewriting the recentfile. Just fetch
1416 the stuff to mirror from several places, bingo. But that's something
1417 that should rather live in a separate package or in rsync directly.
1419 * cronjobs are unsuited because with ntp they would all come at the full
1420 minute and disturb each other. Besides that I'd hate to have a backbone
1421 with more than a few seconds latency.
1423 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1425 * a second rsync server with access control for PAUSE. Port? 873 is the
1426 standard port, let's take 8873.
1428 * if there were a filesystem based on this, it would have a slow access
1429 to inexistent files. It would probably provide wrong readdir (only based
1430 on current content) or also a slow one (based on a recentfile written
1431 after the call). But it would provide fast access to existing files. Or
1432 one would deliberately allow slightly blurred answers based on some
1433 sqlite reflection of the recentfiles.
1435 * todo: write a variant of mirror() that combines two or more
1436 recentfiles and treats them like one
1438 * todo: signal handler to remove the tempfile
1440 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1442 * now that we have the symlink I forgot how it should be used in
1445 * the z loop: add missing files to Z file. Just append them (instead of
1446 prepending). So one guy prepends something from the Y file from time to
1447 time and another guy appends something rather frequently. Collecting
1448 pond. When Y merges into Z, things get epoch and the collecting pond
1449 gets smaller. What exactly are "missing files"?
1451 take note of current epoch of the alpha file, let's call it the
1454 find all files on disk
1456 remove all files registered in the recentworld up to recent-ts
1458 remove all files that have been deleted after recent-ts according to
1461 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1463 * rersyncrecent might be a cronjob with a (locked) state file which
1464 contains things like after and maybe last z sync or such?
1466 rrr-mirror might be an alternative name but how would we justify the
1467 three Rs when there is no Re-Rsync-Recent?
1469 With the --loop parameter it is an endless loop, without it is no loop.
1470 At least this is simple.
1472 * todo: new accssor z-interval specifies how often the Z file is updated
1473 against the filesystem. We probably want no epoch stamp on these
1474 entries. And we want to be able to filter the entries (e.g. no
1475 by-modules and by-category tree)
1477 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1479 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1480 into Z. We do not need the other files filled up with filesystem
1483 * need interface to query for a file in order to NOT call update on
1484 PAUSE a second time within a short time.
1486 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1488 * recommended update interval? Makes no sense, is different for
1495 change-log-default-name: "Todo"