1 2011-01-30 Andreas J. Koenig <andk@cpan.org>
3 * rrr-init and rrr-server need to be used such:
9 Stupid but no showstopper ATM.
11 2010-10-26 Andreas J. Koenig <andk@cpan.org>
13 * forking slave is done, needs more testing.
15 2010-07-06 Andreas J. Koenig <andk@cpan.org>
17 * The two next most important goals are to get a useable rrr-server and
18 a forking slave implementation that uses virtually no memory after the
19 first pass through. Does it need the status file?
21 * rrr-init now being fast we can now remove the alpha alerts. Done.
23 2010-06-16 Andreas J. Koenig <andk@cpan.org>
25 * rrr-init seems to be sssslllloooowwww. At 3:49 I started it for the
26 video tree. After twelve minutes the file has reached a size of 500k. It
27 is locking and rewriting the YAML after every file. Wasn't there an
28 unlocked variant? For "init" one would not expect locked correctness and
29 a rewrite after every item. Given that this tends to become slower with
30 every file it is really unacceptable.
32 Reminds me of rrr-dirtyupdate (the former name of rrr-update). It seems
33 the code was not dirty enough and that was the reason why it was so
34 slow. And rrr-update does not have a --dirty switch, so it looks like we
35 have no way to force dirty and fast operation yet but we need it for
38 Irritating the bootstrapping race: the order {(1) fill initial
39 array, (2) start server} leaves the need for an fsck to cover the time
40 between 1 and 2. Given that regular fsck is needed in any case the price
41 is not high but still doesn't give a warm feeling. Maybe starting the
42 server will by itself do an fsck?
44 2010-06-15 Andreas J. Koenig <andk@cpan.org>
46 * remove the wild alpha alerts and say, it is alpha but parts of it are
47 already working very well or so.
49 2009-12-10 Andreas J. Koenig <andk@cpan.org>
51 * Prominent request for statusfile to safe memory consumption. I'd like
52 to work backwards: get a verification program that verifies how well in
53 sync two nodes are without using frmr itself. Pretend that the forking
54 agent already works and can be interrupted with ^C. Then let it die at
55 random and restart at random and see the memory consumption and the
56 failure rate over time. Compare it with the curve we have now.
58 Of course we also invented the status file as a sharp debugging tool.
59 I'd like to see it in action. See also: F:R:M:Recent::thaw().
61 2009-06-22 Andreas J. Koenig <andk@cpan.org>
63 * Todo: rrr-init to generate some index files so that rrr-server can be
66 2009-05-02 Andreas J. Koenig <andk@cpan.org>
68 * DONE: rename rrr-dirtyupdate to rrr-update and add an option --dirty
69 so we can use it for normal update with a single file. And add a cronjob
70 that runs rrr-fsck so we get a reminder when something is not working
73 2009-04-27 Andreas J. Koenig <andk@cpan.org>
75 * Todo: I still see more link_stat errors than expected. I come back to
76 the former suspicion that we do not do necessary bookkeeping on deleted
77 files so that when we reach the "new" event on a meanwhile deleted file
78 we try to mirror it although we could know better. It's just noise, no
79 harm but noise is irritating.
81 * Todo: more experimenting with runstatusfile to get it official
83 * Todo: make sure $ENV{LANG}="C" when we call rsync because we will
86 rsync: link_stat "/authors/id/T/TH/THINC/DateTime-Format-Flexible-0.02.tar.gz" (in PAUSE) failed: No such file or directory
88 rsync: link_stat "/authors/id/A/AG/AGENT/SSH-Batch-0.001.meta" (in PAUSE) failed: No such file or directory (2)
90 And that ignore_link_stat_error is propagated in sparse_clone.
94 For the record: the bug was that ignore_link_stat_error was lost during
95 sparse_clone. And then the second bug was that it did not default to
96 true. Both are now fixed.
98 2009-04-25 Andreas J. Koenig <andk@cpan.org>
100 * Todo: fill rrr program with life
102 2009-04-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
104 * Minor bug: I see complaints about files not existing like
105 /authors/id/G/GW/GWADEJ/SVG-Sparkline-0.2.0.meta or
106 /authors/id/R/RC/RCAPUTO/POE-1.004.readme
108 In the RECENT files these are delete events but of course also new
109 events in older recent files.
111 I think we do not properly remember deletes when we skip-deletes and so
112 do not filter them out when we later see the "new" event.
114 Yes, but that what ignore_link_stat_error is for, we know about this
115 race condition. We must make ignore_link_stat_error default to true.
119 * bug in Done.pm failing to merge two fields into one when a third is
120 present. FIXED and accompanied with new test.
122 2009-04-24 Andreas J. Koenig <andk@cpan.org>
124 * Todo: keep the index clean and run some kind of nondestructive fsck 4
127 * who is our backbone?
129 http://cpan.cpantesters.org/authors/02STAMP # barbie
130 http://cpan.solfo.com/authors/02STAMP # abh
131 (http|ftp)://theoryx5.uwinnipeg.ca/pub/CPAN/ # rkobes
133 2009-04-17 Andreas J. Koenig <andk@cpan.org>
135 * the changelog helper I included in the Makefile release memo does not
136 show tags and is much less useful than I thought. gitk is probably much
139 2009-04-16 Andreas J. Koenig <andk@cpan.org>
143 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
144 #### Ydr_.yaml ... DONE
145 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
146 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
147 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
148 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
149 #### gone already at cpan-pause.pl line 0
151 Running without skip-deletes now but cannot reproduce.
153 2009-04-15 Andreas J. Koenig <andk@cpan.org>
155 * consider state directory such we can restart after a ^C or come again
158 Reuse _runstatusfile: write atomically, see how we can use the dump to
159 start running again. Do not leak and remove globs before writing. Do not
160 only dump one rf, dump the whole complex.
162 * consider the effect of ^C. Is it safe?
164 * want to use Perl::Version to up the versions on every release. Only if
165 things have changed, of course. Does this actually work?
167 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
168 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
170 Must do the bookkeeping myself when this should happen, maybe based on git?
172 I think I prefer the same version number for all pm files. But one also
173 needs a routine for setting them. Ahh:
175 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
181 cpan[2]> m /Mirror::Recent/
182 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
183 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
184 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
185 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
188 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
190 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
192 I just tried MM 6.5102 and it again produces only a provides for Recent,
193 not for the other modules.
195 My own fault. FIXED in the Makefile.PL.
197 2009-04-13 Andreas J. Koenig <andk@cpan.org>
199 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
201 * is it true (as stated at
202 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
203 that rsync can lead to errors when upstream changes before client sync
204 has completed? Can we see the error when we run this on upstream:
207 use Time::HiRes qw(time);
209 open my $fh, ">", "changingfile.txt.new" or die;
210 print $fh (time."\n") x 1000000;
212 rename "changingfile.txt.new", "changingfile.txt" or die;
216 And on the receiving end:
219 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
223 I cannot get it to fail. Apparently it is sufficient when upstream
224 always writes atomically (which of course is mandatory)?
226 SENT to mailing list Apr 28.
230 > https://fedorahosted.org/InstantMirror/
231 > https://www.redhat.com/mailman/listinfo/instantmirror-list
232 > irc.freenode.net channel #instantmirror
234 A cool name for a project. Inspires me to write a few sentences about
235 bittorrent's role in the grand picture.
237 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
238 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
240 2009-04-12 Andreas J. Koenig <andk@cpan.org>
242 * Interesting last minute bug during real download testing: the output
243 isn't as pretty anymore, it seems that more work is being done than
246 Introducing a highlevel _alluptodate method helped a lot in debugging
249 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
250 different on 1h file than on the other files. This lead to frequent
251 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
254 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
256 * Todo: try if we can be faster with a native float package. I'm really
257 glad to have a machine exposing perl floating point bugs, it forced me
258 to retract this item from the changes file: "several speedups in the
259 fakebigfloat code" before I get cpan testers fails.
261 2009-04-11 Andreas J. Koenig <andk@cpan.org>
263 * Bug: when the dirtymark changes upstream, then the downstream server
264 notices it immediately and mirrors the "1h" file but then leaves the
265 rmirror loop and continues with "6h" on the next round through. This
266 behaviour goes on until it reaches the Z file. Because we eat the Z file
267 piecemeal we leave the loop after a while but after that everybody seems
268 to have forgotten that there is some work left to be done. Bad, bad,
271 ETOOCONFUSING better name?
273 Recentfile::get_remote_recentfile_as_tempfile OK
274 Recentfile::resolve_recentfilename split_rfilename DONE
276 Recent::_principal_recentfile_object _principal_recentfile_fromremote
277 Recent::principal_recentfile OK
278 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
280 some void, some not, some not void but called in void context.
282 2009-04-10 Andreas J. Koenig <andk@cpan.org>
284 * Bug? When the dirtymark changes then the second tier hosts quickly
285 reset their DONE state and restart mirroring. But then they mirror
286 potentially outdated index files with inconsistent dirtymark because the
287 first tier box may be a bit behind on the large recentfiles.
289 This means we need a brake when iterating over the recentfiles that
290 refuses to use a recentfile with the wrong dirtymark.
292 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
294 * install the bin/* files and talk about rrr-overview somewhere.
296 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
297 created outside the target tree.
299 * Barbie asks for better logging capabilities
301 2009-03-29 Andreas J. Koenig <andk@cpan.org>
303 * known bug: cannot configure the Recentfile objects to keep delete
304 events. How to call the accessor for that? keep_delete_objects_forever?
306 Sounds acceptable. FIXED (but untested).
308 * known bug: keeps temporary index files lying around.
310 2009-03-24 Andreas J. Koenig <andk@cpan.org>
312 * Some equivalent for
314 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
318 2009-03-22 Andreas J. Koenig <andk@cpan.org>
320 * k81 is again client of k75 with different parameters.
322 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
323 about it all the time and nobody reacts accordingly; (2) lots of temp
324 files get created and not removed; culprit the new call to
325 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
326 caller has to remove the tempfile after use.
328 Need the drawing board.
330 PARTLY FIXED: retracting the idea to call
331 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
332 and unseed such that uptodate now means "we have mirrored this
333 recentfile's bunch of files" and seeded means "somebody believes the
334 index file for this rf needs be refreshed". This seems to work now.
336 Still collecting temp files that nobody cares to depollute.
338 2009-03-21 Andreas J. Koenig <andk@cpan.org>
340 * Bug?: should it be harder than it is atm to set the timestamp to the
343 * bug with native integers:
347 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
351 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
354 Native integer broke when native math was turned off. FIXED
356 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
357 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
360 Yes, records out of order:
363 447004 epoch: 995885533
364 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
367 447008 epoch: 995890358
368 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
371 447012 epoch: 995892221
372 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
375 FIXED with sanity check and later with the integer fix.
377 * Bug: want the index files in a .recent directory
379 * Bug: lots of dot files are not deleted in time
381 * possible test case: can a delete change the timestamp? This would
382 probably break the order of events.
384 2009-03-20 Andreas J. Koenig <andk@cpan.org>
386 * 1233701831.34486 what's so special about this number/string? It
388 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
389 atm lives in Y,Q, and Z.
391 It is the first entry after id/--skip-locking which has timestamp
392 1234164228.11325 which represents a file that doesn't exist anymore.
396 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
397 _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
398 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
399 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
400 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
401 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
402 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
403 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
404 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
405 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
406 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
407 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
408 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
410 FIXED, it was the "--skip-locking" file where manual intervention was participating
412 * bug on the mirroring slave: when the dirtymark gets increased we
413 probably do not reset the done intervals. The mirrorer stays within
414 tight bounds where it tries to sync with upstream and never seems to
415 finish. In the debugging state file I see lots of identical intervals
416 that do not get collapsed. When I restart the mirrorer it dies with:
418 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
419 _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
420 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
421 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
422 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
423 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
424 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
425 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
426 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
427 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
429 and debugging stands at
435 epoch: 1237400802.5789
436 path: id/J/JO/JOHND/CHECKSUMS
439 epoch: 1237400807.97514
440 path: id/M/MI/MIYAGAWA/CHECKSUMS
443 epoch: 1237400817.94363
444 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
452 and it is reproducable.
454 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
455 pause has a fresh one.
457 FIXED, it was the third anded term in each of the ifs in the IV block in
458 _register_one: with that we make sure that we do not stamp on valuable
461 2009-03-17 Andreas J. Koenig <andk@cpan.org>
463 * done: verified the existence of the floating point bug in bleadperl
464 and verified that switching from YAML::Syck to YAML::XS does not resolve
467 BTW, the switch was doable with
469 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
471 and should be considered as a separate TODO
473 * todo: integrate a dirty update with two aggregate calls before
474 unlocking for frictionless dirtying DONE
476 * todo: start the second rsync daemon on pause DONE
478 * todo: move index files to .recent: this cannot simply be done by
479 setting filenameroot to .recent/RECENT. Other parts of the modules rely
480 on the fact that dirname(recentfile) is the root of the mirrored tree.
482 2009-03-16 Andreas J. Koenig <andk@cpan.org>
484 * What was the resolution of the mirror.pl delete hook bug? Do we call
485 the delete hook when pause removes a file from MUIR?
487 * Today on pause: Updating 2a13fba..29f284d and installing it for
488 /usr/local/perl-5.10.0{,-RC2}
490 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
491 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
493 2009-03-15 Andreas J. Koenig <andk@cpan.org>
495 * currently recent_events has the side effect of setting dirtymark
496 because it forces a read on the file. That should be transparent, so
497 that the dirtymark call always forces a cache-able(?) read.
499 * The bug below is -- after a lot of trying -- not reproducible on a
500 small script, only in the large test script. The closest to the output
506 my $x = "01237123229.8814";
511 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
513 $l = "1237123231.22458";
514 $r = "1237123231.22458";
516 Devel::Peek::Dump $l;
517 Devel::Peek::Dump $r;
518 Devel::Peek::Dump $x = $l <=> $r;
522 The checked in state at c404a85 fails the test with my
523 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
524 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
525 in the test script always consistent.
527 * Todo: write a test that inserts a second dirty file with an already
528 existing timestamp. DONE
530 * Bug in perl 5.10 on my 64bit box:
532 DB<98> Devel::Peek::Dump $l
533 SV = PVMG(0x19e0450) at 0x142a550
535 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
537 NV = 1237123231.22458
538 PV = 0x194ce70 "1237123231.22458"\0
542 DB<99> Devel::Peek::Dump $r
543 SV = PVMG(0x19e0240) at 0x142a3e8
545 FLAGS = (PADMY,POK,pPOK)
548 PV = 0x19ff900 "1237123231.22458"\0
552 DB<100> Devel::Peek::Dump $l <=> $r
553 SV = IV(0x19ea6e8) at 0x19ea6f0
555 FLAGS = (PADTMP,IOK,pIOK)
558 DB<101> Devel::Peek::Dump $l
559 SV = PVMG(0x19e0450) at 0x142a550
561 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
563 NV = 1237123231.22458
564 PV = 0x194ce70 "1237123231.22458"\0
568 DB<102> Devel::Peek::Dump $r
569 SV = PVMG(0x19e0240) at 0x142a3e8
571 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
573 NV = 1237123231.22458
574 PV = 0x19ff900 "1237123231.22458"\0
578 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
580 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
582 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
584 * $obj->merge ($other) needs to learn about equal epoch which may happen
585 since dirty_epoch intruded.
587 * Wontfix anytime soon: I think we currently do not support mkdir. Only
590 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
592 * Todo: continue working on update(...,$dirty_epoch). It must be
593 followed by a fast_aggregate! DONE
595 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
597 * maybe we need a closest_entry or fitting_interval or something like
598 that. We want to merge an event into the middle of some recentfile.
599 First we do not know which file, then we do not know where to lock,
600 where to enter the new item, when and where to correct the dirtymark.
602 So my thought is we should first find which file.
604 Another part of my brain answers: what would happen if we would enter
605 the new file into the smallest file just like an ordinary new event,
606 just as an old event?
608 (1) we would write a duplicate timestamp? No, this would be easy to
611 (2) we would make the file large quickly? Yes, but so what? We are
612 changing the dirtymark, so are willing to disturb the downstream hosts.
614 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
616 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
620 https://rt.cpan.org/Ticket/Display.html?id=41199
622 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
624 Strangest output being HAKANARDO who managed to upload
626 Here is a better oneliner that includes also the first line of each
629 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
631 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
633 * memory leak in the syncher? It currently weighs 100M.
637 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
640 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
642 * lookup by epoch and by path and use this ability on the pause to never
643 again register a file twice that doesn't need it. Let's call it
646 * after the dirtymark is done: fill up recentfiles with fake (historic)
647 entries; fill up with individual corrections; algorithm maybe to be done
648 with bigfloat so that we can always place something in the middle
649 between two entries. Before we must switch to bigfloat we could try to
650 use Data::Float::nextup to get the.
652 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
655 * dirtymark now lives in Recentfile, needs to be used in rmirror.
657 * find out why the downloader died after a couple of hours without a net
658 connection. Write a test that survives the not-existence of the other
661 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
663 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
664 it makes less sense with tight coupling for secondary files. But asking
665 the server what the current epoch is might be cheaper on HTTP than on
666 rsync. (Needs to be evaluated)
668 * remove the 0.00 from the verbose overview in the Merged column in the
671 * write tests that expose the problems of the last few days: cascading
672 client/server roles, tight coupling for secondary RFs, deletes after
675 * Some day we might want to have policy options for the slave:
676 tight/loose/no coupling with upstream for secondary RFs. tight is what
677 we have now. loose would wait until a gap occurs that can be closed.
679 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
681 * revisit all $rfs->[$i+1] places if they now make sense still
683 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
685 * another bug is the fact that the mirror command deletes files before
686 it unhides the index file, thus confusing downstream slaves. We must not
687 delete before unhiding and must delete after unhiding. FIXED.
689 * new complication about the slave that is playing a server role.
690 Currently we mirror from newest to oldest with a hidden temporary file
691 as index. And when one file is finished, we unhide the index file.
692 Imagine the cascading server/slave is dead for a day. It then starts
693 mirroring again with the freshest thing and unhides the freshest index
694 file when it has worked through it. In that moment it exposes a time
695 hole. Because it now works on the second recentfile which is still
698 We currently do nothing special to converge after such a drop out. At
699 least not intentionally and robustly and thought through.
701 The algorithm we use to seed the next file needs quite a lot of more
702 robustness than it currently has. Something to do with looking at the
703 merged element of the next rf and when it has dropped off, we seed
704 immediately. And if it ramains dropped off, we seed again, of course.
706 Nope, looking from smaller to larger RFS we look at the merged element
707 of this RF and at the minmax/max element of the next RF. If that
708 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
711 And we need a public accessor seed and unseed or seeded. But not the mix
712 of public and private stuff that then is used behind the back.
714 And then the secondary* stuff must go.
716 And we must understand what the impact is on the DONE system. Can it go
717 unnoticed that there was a hole? And could the DONE system have decided
718 the hole is covered? This should be testable with three directories where
719 the middle stops working for a while. Done->merge is suspicious, we must
720 stop it from merging non-conflatable neighbors due to broken continuity.
724 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
726 * Slaven suggests to have the current epoch or the whole current
727 recentfile available from the HTTP server and take it away with
728 keepalive. This direction goes the granularity down to subseconds.
730 We might want to rewrite everything to factor out transport and allow
731 the whole thing to run via HTTP.
733 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
735 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
738 * smoker on k81 fetching from k75 to verify cascading works. See
739 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
741 * maybe the loop should wait for CHECKSUMS file after every upload. And
742 CPAN.pm needs to deal with timestamps in the future.
744 * do not forget the dirtymark!
746 Text: have a new flag on recentfiles with the meaning: if this
747 changes, you're required to run a full rsync over all the files. The
748 reason why we set it would probably be: some foul happened. we injected
749 files in arbitrary places or didn't inject them although they changed.
750 The content of the flag? Timestamp? The relation between the
751 recentfiles would have to be inheritance from the principal, because any
752 out of band changes would soon later propagate to the next recentfile.
754 By upping the flag often one can easily ruin the slaves.
756 last out of band change? dirtymark?
758 Anyway, this implies that we read a potentially existing recentfile
761 And it implies that we have an eventloop that keeps us busy in 2-3
762 cycles, one for current stuff (tight loop) and one for the recentfiles
763 (cascade when principal has changed), one for the old stuff after a
766 And it implies that the out-of-band change in any of the recentfiles
767 must have a lock on the principal file and there is the place to set the
770 * start a FAQ, especially quick start guide questions. Also to aid those
771 problematic areas where we have no good solution, like the "links"
774 * wish feedback when we are slow.
778 * Remove a few DEBUG statements.
780 * The multiple-rrr way of doing things needs a new option to rmirror,
781 like piecemeal or so. Not urgent because after the first pass through,
782 things run smoothely. It's only ugly during the first pass.
784 * I have the suspicion that the code is broken that decides if the
785 neighboring RF needs to be seeded. I fear when too much time has gone
786 between two calls (in our case more than one hour), it would not seed
787 the neighbor. Of course this will never be noticed, so we need a good
790 * local/localroot confusion: I currently pass both options but one must
793 * accounts for early birds on PAUSE rsync daemon.
795 * hardcoded 20 seconds
797 * who mirrors the index? DOING now.
799 * which CPAN mirrors offer rsync?
801 * visit all XXX, visit all _float places
803 * rename the pathdb stuff, it's too confusing. No idea how.
805 * rrr-inotify, backpan, rrr-register
807 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
809 * current bugs: the pathdb seems to get no reset, the seeding of the
810 secondaryttl stuff seems not to have an effect. Have helped myself with
811 a rand(10), need to fix this back. So not checked in. Does the rand
814 The rand thing helps. The secondaryttl stuff was in the wrong line,
817 The pathdb stuff was because I called either _pathdb or __pathdb on the
818 wrong object. FIXED now.
820 * It's not so beautiful if we never fetch the recentfiles that are not
821 the principal, even if this is correct behaviour. We really do not need
822 them after we have fetched the whole content.
824 OK, we want a switch for that: secondaryttl DONE
826 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
828 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
829 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
830 it lets 40 follow. FIXED
832 * See that the long running process really only updates the principal
833 file unless it has missed a timespan during which something happened. If
834 nothing happened, it must notice even when it misses the timespan. DONE
836 * we must throw away the pathdb when we have reached the end of Z. From
837 that moment we can have a very small pathdb because the only reason for
838 a pathdb is that we know to ignore old records in old files. We won't
839 need this pathdb again before the next full pass over the data is
840 necessary and then we will rebuild it as we go along. DONE
842 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
844 * I think, Done::register_one is doing wrong in that it does not
845 conflate neighboring pieces. The covered() method cannot do this because
846 it has no recent_events array at hand. But register_one has it and could
847 do it and for some reason misses to do it (sometimes).
849 This means that the three tests I just wrote can probably not survive
850 because they test with an already broken Done structure.
852 The art now is to detect how it happens, then to reproduce, then write a
855 So from the logfile this is what happens: we have a good interval with
856 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
857 top of it. Locally we now mirror F2 and open a new done interval for it.
858 Then we mirror F1 but this time with the timestamp T1b. And when we then
859 try to close the gap, we do not find T1 but instead something older. We
860 should gladly accept this older piece and this would fix this bug.
864 * bug to fix: when the 1h file changes while rmirror is running, we do
865 correctly sync the new files but never switch to the 6h file but rather
866 stay in a rather quick loop that fetches the 1h file again and again.
868 Is it possible that we initialize a new object? Or does
869 get_remote_recentfile_as_tempfile overwrite something in myself?
871 Want a new option: _runstatusfile => $file which frequently dumps the
872 state of all recentfiles to a file.
876 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
878 * Todo: now teach update to verify the timestamp is about to write
879 against the previous and use _increase_a_bit if it doesn't comply with
880 strict monotony. DONE
882 * The problem of rounding. So far perl's default precision was
883 sufficient. One day it won't be. FakeFloat has an easy job when it is
884 only reading and other machines have written correctly. But when we want
885 to write a floating point number that is a bit larger than the other
886 one, then we need our own idea of precision.
888 Slaven said: just append a "1". This might be going towards the end of
889 usability too quickly. I'd like something that actually uses the decimal
890 system. Well, appending a 1 also does this but...
892 E.g. we have 1.0. nextup on this architecture is starting with
893 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
894 taken the 1.0000000000000003 and the next user comes and the time tells
895 him 1.0 again. He has to beat my number without stepping over the
896 nextup. This is much less space than I had when I chose 1,2,3.
898 What is also irritating is that nextup is architecture dependent. The
899 128 bit guy must choose very long numbers to fit in between whereas the
900 other one with 16 bit uses larger steps. But then the algorithm is the
901 same for both, so that would be a nice thing.
903 I see two situation where we need this. One is when Time::HiRes returns
904 us a value that is <= the last entry in our recentfile. In this case
905 (let's call it the end-case) we must fill the region between that number
906 and the next higher native floating point number. The other is when we
907 inject an old file into an old recentfile (we would then also set a new
908 dirtymark). We find the integer value already taken and need a slightly
909 different one (let's call it the middle-case). The difference between
910 the two situations is that the next user will want to find something
911 higher than my number in the end-case and something lower than my number
914 So I suggest we give the function both a value and an upper bound and it
915 calculates us a primitive middle. The upper bound in the middle-case is
916 the next integer. The upper bound on the end-case is the nextup floating
917 point number. But the latter poses another problem: if we have occupied
918 the middle m between x and nextup(x), then the nextup(m) will probably
919 not be the same as nextup(x) because some rounding will take place
920 before the nextup is calculated and when the rounding reaches the
921 nextup(x), we will end up at nextup(nextup(x)).
923 So we really need to consider the nextup and the nextdown from there and
924 then the middle and that's the number we may approach asymptotically.
927 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
929 * consider deprecating the use of RECENT.recent as a symlink. It turns
930 out to need extra hoops with the rsync options and just isn't worth it.
931 Or maybe these extra hoops are needed anyway for the rest of the tree?
932 Nope, can't be the case because not all filesystems support symlinks.
934 But before doing the large step, I'll deprecate the call of
935 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
936 only introduced to resolve RECENT.recent and complicates the routine far
937 beyond what it deserves.
939 DONE. Won't deprecate RECENT.recent, just moved its handling to the
942 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
944 * I think it's a bug that the rsync_option links must be set to true in
945 order to support RECENT.recent and that nobody cares to set it
946 automatically. Similar for ignore_link_stat_errors.
948 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
950 * Todo: find all todos together and make a plan what is missing for a
953 - verifytree or something like that. fsck maybe.
955 - rersyncrecent, the script itself? What it do?
957 - a way to only mirror the recentfiles without mirroring the whole
958 remote system such that people can decide to mirror only partially see
959 also 2008-08-30. .shadow-xxx directory? this also needed for a
960 filesystem that is still incomplete and might need the mirrorfiles for
963 - long living objects that mirror again and again. Inject something
964 into ta, see how it goes over to tb.
966 - how do we continue filling up the DONE system when we use an object
967 for the second time? "fully covered" and "uptodate" or new terminology.
969 - overview called on the wrong file should be understandable
971 - the meta data field that must change when we fake something up so that
972 the downstream people know they have to re-fetch everything.
974 - how tolerant are we against missing files upstream? how do we keep
975 track? there are legitimate cases where we did read upstream index right
976 before a file got deleted there and then find that file as new and want
977 it. There are other cases that are not self healing and must be tracked
980 - how, exactly, do we have to deal with deletes? With rsync errors?
982 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
983 authors) failed: No such file or directory (2)
985 The file above is a delete in 1h and a new in file 1M and the
986 delete in the locally running rmirror did not get propagated to the 1M
987 object. Bug. And the consequence is a standstill.
989 It seems that a slave that works with a file below the principal needs
990 to merge things all the way up to get rid of later deletes. Or keep
991 track of all deletes and skip them later. So we need a trackdeletes.pm
992 similar to the done.pm?
994 see also 2008-08-20 about spurious deletes that really have no add
995 counterpart and yet they are not wrong.
997 - consider the effect when resyncing the recentfile takes longer than
998 the time per loop. Then we never rsync any file. We need to diagnose
999 that and force an increase of that loop time. But when we later are fast
1000 enough again because the net has recovered, then we need to switch back
1001 to original parameters. ERm, no, it's enough to keep syncing at least
1002 one file before refetching an index file.
1004 - remember to verify that no temp files are left lying around and the
1007 - status file for not long running jobs that want to track upstream with
1010 - revisit all XXX _float areas and study Sub::Exporter DONE
1012 - persistent DB even though we just said we do not need it. Just for
1013 extended capabilities and time savings when, for example, upstream
1014 announces a reset and we get new recentfiles and could then limit
1015 ourselves to a subset of files (those that have a changed epoch) in a
1016 first pass and would only then do the loop to verify the rest. Or
1019 * Todo: aggregate files should know their feed and finding the principal
1020 should be done stepwise. (?)
1022 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
1023 left locks around. DONE
1025 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1027 * maybe extend the _overview so that it always says if and where the
1028 last file is in the next file and where the next event in the next rf
1029 would lie. No, don't like this anymore. REJECT
1031 * take the two new redundant tests out again, only the third must
1034 * Todo: add a sanity check if the merged structure is really pointing to
1035 a different rf and that this different rf is larger. DONE
1037 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1039 * now test, if they are overlapping. And test if there is a file in the
1040 next rf that would fit into this rf's interval.
1042 1h 1222324012.8474 1222322541.7963 0.4086
1043 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
1044 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
1045 1W 1222313218.3626 1221708477.5829 167.9835
1047 I suspect that somebody writes a merged timestamp without having merged
1048 and then somebody else relies on it.
1050 If aggregate is running, the intervals must not be extravagated, if it
1051 is not running, there must not be bounds, the total number of events in
1052 the system must be counted and must be controlled throughout the tests.
1053 That the test required the additional update was probably nonsense,
1054 because aggregate can cut pieces too. FIXED & DONE
1056 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1058 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
1061 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1063 * Most apparent bug at the moment is that the recentfiles are fetched
1064 too often. Only the principal should be fetched and if it has not
1065 changed, the others should not be refetched. ATM I must admit that I'm
1066 happy that we refetch more often than needed because I can more easily
1067 fix bugs while the thing is running.
1069 * Let's say, 1220474966.19501 is a timestamp of a file that is already
1070 done but the done system does not know about it. The reason for the
1071 failure is not known and we never reach the status uptodate because of
1072 this. We must get over it.
1074 Later it turns out that the origin server had a bug somewhere.
1075 1220474966.19042 came after 1220474966.19501. Or better: it was in the
1076 array of the recentfile one position above. The bug was my own.
1078 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1080 * There is the race condition where the server does a delete and the
1081 slave does not yet know and then tries to download it because he sees
1082 the new. So for this time window we must be more tolerant against
1083 failure. If we cannot download a file, we should just skip it and should
1084 not retry immediately. The whole system should discover the lost thing
1085 later. Keeping track with the DONE system should really be a no brainer.
1087 But there is something more: the whole filesystem is a database and the
1088 recentfiles are one possible representation of it. It's a pretty useful
1089 representation I think that's why I have implemented something around
1090 it. But for strictly local operation it has little value. For local
1091 operation we would much rather have a database. So we would enter every
1092 recentfile reading and every rsync operation and for every file the last
1093 state change and what it leads to. Then we would always ignore older
1094 records without the efforts involved with recentfiles.
1096 The database would have: path,recentepoch,rsyncedon,deletedon
1098 Oh well, not yet clear where this leads to.
1100 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1102 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
1103 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
1105 * statusfile, maybe only on demand, alone to have a sharp debugging
1106 tool. It is locked and all recentfiles dump themselves into it and we
1107 can build a viewer that lets us know where we stand and what's inside.
1109 * remember: only the principal recentfile needs expiration, all others
1110 shall be expired by principal if it discovers that something has move
1113 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1115 * Always check if we stringify to a higher value than in the entry
1118 * And in covered make an additional check if we would be able to see a
1119 numerical difference between the two numbers and if we can't then switch
1120 to a different, more expensive algorithm. Do not want to be caught by
1121 floating surprises. DONE
1123 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1125 * caching has several aspects here: we can cache the interval of the
1126 recentfile which only will change when the mtime of the file changes. We
1127 must re-mirror the recentfile when its ttl has expired. Does have_read
1128 tell you anything? It counts nothing at all. Only the mtime is
1129 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
1130 thing the have_mirrored because it is unrelated to the mtime.
1132 * Robustness of floating point calculations! I always thought that the
1133 string calculated by the origin server for the floating representation
1134 of the epoch time is just a string. When we convert it to a number and
1135 later back to a string, the other computer might come to a different
1136 conclusion. This must not happen, we want to preserve it under any
1137 circumstances. I will have to write tests with overlong sequences that
1138 get lost in arithmetic and must see if all still works well. DONE
1140 But one fragile point remains: if one host considers a>b and the other
1141 one considers them == but no eq. To prevent this, we must probably do
1142 some extra homework. DONE
1144 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1146 * the concept of tracking DONE needs an object per recentfile that has
1147 something like these methods:
1149 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1151 covered() register() covered()
1153 The unclear thing is how we translate points in time into intervals. We
1154 could pass a reference to the current recent_events array when running
1155 we_have(xxx) and let the DONE object iterate over it such that it only
1156 has to store a list of intervals that can melt into each other. Ah, even
1157 passing the list together with a list of indexes seems feasiable.
1159 Or maybe ask for the inverted list?
1161 Whenever the complete array is covered by the interval we say we are
1162 fully covered and if the recentfile is not expired, we are uptodate.
1164 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1166 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1168 * need a way to "return" the next entry after the end of a list. When
1169 the caller says "before" or "after" we would like to know if he could
1170 cover that interval/threshold or not because this influences the effect
1171 of a newer timestamp of that recentfile. DONE with $opt{info}.
1173 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1175 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1177 loop implies tixu (?). Nope, something like --statefile decides. Per
1180 T test, I init, X including removals, U nodirtymark
1182 So we have no concept of dirtymarks, we only trust that since we are
1183 running we have observed everything steadily. But people will not let
1184 this program run forever so we must consider both startup penalty and
1185 book keeping for later runs. We keep this for later. For now we write a
1186 long running mirror that merges several intervals.
1188 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1190 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1193 * rersyncrecent, the script: default to one week. The name of the switch
1194 is --after. Other switches? --loop!
1196 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1198 * need a switch --skip-deletes (?)
1200 * need a switch --enduser that tells us that the whole tempfile
1201 discipline is not needed when there is no downstream user. (?)
1203 Without this switch we cannot have a reasonable recent.pl that just
1204 displays the recent additions. Either we accept to download everything.
1205 Or we download temporary files without the typical rsync protocol
1208 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1209 File::Temp::tempdir, this might be a win.
1211 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1213 * apropos missing: we have no push, we never know the downstream
1214 servers. People who know their downstream hosts and want to ascertain
1215 something will want additional methods we have never thought about, like
1216 update or delete a certain file.
1218 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1220 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1221 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1224 * rsync options: it's a bit of a pain that we usually need several rsync
1225 options, like compress, links, times, checksum and that there is no
1226 reasonable default except the original rsync default. I think wee can
1227 safely assume that the rsync options are shared between all recentfile
1228 instances within one recent tree.
1230 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1232 * deletes: if a delete follows an add quickly enough it may happen that
1233 a downstream mirror did not see the add at all! It seems this needs to
1234 be mentioned somewhere. The point here is that even if the downstream is
1235 never missing the principal timeframe it may encounter a "delete" that
1236 has no complimentary "add" anywhere.
1238 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1240 * I suspect the treat of metadata is incorrect during read or something.
1241 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1242 contained more than 6 hours worth of data. At 06:08 we merged into the
1243 1d file. We need to take snapshots of the 6h file over the course of an
1244 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1247 Much worse: watching the 1h file: right at the moment (at 06:35) it
1248 covers 1218867584-1219120397 which is 70 hours.
1250 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1251 08:19:44 2008, that is when I checked out last time, so it seems to be
1252 aggregating and never truncating?
1254 No, correct is: it is never truncating; but wrong is: it is aggregating.
1255 It does receive a lot of events from time to time from a larger file.
1256 Somehow a large file gets merged into the small one and because the
1257 "meta/merged" attribute is missing, nobody is paying attention. I
1258 believe that I can fix this by making sure that metadata are honoured
1259 during read. DONE and test adjusted.
1261 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1263 * grand renaming plan
1265 remotebase => remoteroot to fit well with localroot DONE
1266 local_path() => localroot seems to me should already work DONE
1267 recentfile_basename => rfilename no need to stress it has no slash DONE
1269 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1270 kern? It would anyway need a deprecation cycle because it is an
1271 important constructor.
1273 * I like the portability that Data::Serializer brings us but the price
1274 is that some day we might find out that it is slowing us a bit. We'll
1277 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1279 * should we not enter the interval of the principal (or the interval of
1280 the merging file?) in every aggregated/merged file?
1282 * we should aim at a first release and give up on thinking about
1283 sanitizing stuff and zloop. Let's just admit that a full traditional
1284 rsync is the only available sanitizer ATM. Otherwise it's complicated
1285 stuff: sanitizing on the origin server, sanitizing on the slaves,
1286 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1287 get the basics out before this becomes a major cause for mess.
1289 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1291 * On OSes not supporting symlinks we expect that RECENT.recent contains
1292 the contents of the principal recentfile. Actually this is identical on
1293 systems supporting symlinks. Simple, what follows from that is that we
1294 need to keep the serializer in the metadata because we cannot read it
1295 from the filename, doesn't it? Of course not. It's a chicken and egg
1296 problem. This leaves us with the problem to actually parse the
1297 serialized data to find out in which format it is. So who can do the 4
1298 or 5 magics we wanted to support? File::LibMagic?
1300 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1302 * remotebase and recentfile_basename are ugly names. Now that we need a
1303 word for the shortest/principal/driving recentfile too we should do
1306 localroot is good. rfile is good. local_path() is bad, local_path($path)
1307 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1310 Up to now remotebase was the string that described the remote root
1311 directory in rsync notation, like pause.perl.org::authors. And
1312 recentfile_basename was "RECENT-1h.yaml".
1314 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1316 * The test that was added in today's checkin is a good start for a test
1317 of rmirror. We should have more methods in Recent.pm: verify,
1318 addmissingfiles. We should verify the current tree, then rmirror it and
1319 then verifytree the copy. We could then add some arbitrary file and let
1320 it be discovered by addmissingfiles, then rmirror again and then
1321 verifytree the copy again.
1323 Then we could start stealing from csync2 sqlite database [no port to
1324 OSX!] and fill a local DB. And methods to compare the database with the
1325 recentfiles. Our strength is that in principle we could maintain state
1326 with a single float. We have synced up to 1234567890.123456. If the Z
1327 file does not add new files all we have to do is mirror the new ones and
1330 This makes it clear that we should extend current protocol and declare
1331 that we cheat when we add files too late, just to help the other end
1332 keeping track. Ah yes, that's what was meant when zloop was mentioned
1335 Maybe need to revisit File::Mirror to help me with this task.
1337 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1339 * There must be an allow-me-to-truncate flag in every recentfile.
1340 Without it one could construct a sequence of updates winning the locking
1341 battle against the aggregator. Only if an aggregator has managed to
1342 merge data over to the next level, truncating can be allowed. DONE with
1345 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1347 * We should probably guarantee that no duplicates enter the aggregator
1350 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1352 * To get merge operation faster would need a good benchmark test. What
1353 02 spits out isn't reliable enough and is dominated by many other
1356 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1357 Author: Andreas J. Koenig <k@k75.(none)>
1358 Date: Sat Aug 2 07:58:04 2008 +0200
1362 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1363 Author: Andreas J. Koenig <k@k75.(none)>
1364 Date: Sat Aug 2 11:40:29 2008 +0200
1366 there was a lot of trying but the effect is hardly measurable with
1369 * overhead of connecting seems high. When setting
1370 max_files_per_connection to 1 we see that.
1372 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1374 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1376 25138.2215600014/3600 = 6.98283932222261
1378 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1379 be a bug in the aggregator. FIXED
1381 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1383 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1384 we have it, can we mirror it, mirror it!
1386 I fear this needs a new class which might be called
1387 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1388 and treat them as an entity. I realize that a single recentfile may be
1389 sufficient for certain tasks and that it is handy for the low level
1390 programmer but it is not nice to use. If there is a delete in the 1h
1391 file then the 6h file still contains it. Seekers of the best information
1392 need to combine at least some of the recentfiles most of the time.
1394 There is the place for the Z loop!
1396 But the combination is something to collect in a database, isn't it. Did
1397 csync2 just harrumph?
1399 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1401 * it just occurred to me that hosts in the same mirroring pool could
1402 help out each other even without rewriting the recentfile. Just fetch
1403 the stuff to mirror from several places, bingo. But that's something
1404 that should rather live in a separate package or in rsync directly.
1406 * cronjobs are unsuited because with ntp they would all come at the full
1407 minute and disturb each other. Besides that I'd hate to have a backbone
1408 with more than a few seconds latency.
1410 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1412 * a second rsync server with access control for PAUSE. Port? 873 is the
1413 standard port, let's take 8873.
1415 * if there were a filesystem based on this, it would have a slow access
1416 to inexistent files. It would probably provide wrong readdir (only based
1417 on current content) or also a slow one (based on a recentfile written
1418 after the call). But it would provide fast access to existing files. Or
1419 one would deliberately allow slightly blurred answers based on some
1420 sqlite reflection of the recentfiles.
1422 * todo: write a variant of mirror() that combines two or more
1423 recentfiles and treats them like one
1425 * todo: signal handler to remove the tempfile
1427 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1429 * now that we have the symlink I forgot how it should be used in
1432 * the z loop: add missing files to Z file. Just append them (instead of
1433 prepending). So one guy prepends something from the Y file from time to
1434 time and another guy appends something rather frequently. Collecting
1435 pond. When Y merges into Z, things get epoch and the collecting pond
1436 gets smaller. What exactly are "missing files"?
1438 take note of current epoch of the alpha file, let's call it the
1441 find all files on disk
1443 remove all files registered in the recentworld up to recent-ts
1445 remove all files that have been deleted after recent-ts according to
1448 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1450 * rersyncrecent might be a cronjob with a (locked) state file which
1451 contains things like after and maybe last z sync or such?
1453 rrr-mirror might be an alternative name but how would we justify the
1454 three Rs when there is no Re-Rsync-Recent?
1456 With the --loop parameter it is an endless loop, without it is no loop.
1457 At least this is simple.
1459 * todo: new accssor z-interval specifies how often the Z file is updated
1460 against the filesystem. We probably want no epoch stamp on these
1461 entries. And we want to be able to filter the entries (e.g. no
1462 by-modules and by-category tree)
1464 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1466 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1467 into Z. We do not need the other files filled up with filesystem
1470 * need interface to query for a file in order to NOT call update on
1471 PAUSE a second time within a short time.
1473 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1475 * recommended update interval? Makes no sense, is different for
1482 change-log-default-name: "Todo"