1 2009-04-16 Andreas J. Koenig <andk@cpan.org>
5 #### Sync 1239828608 (1/1/Z) temp .../authors/.FRMRecent-RECENT-Z.yaml-
6 #### Ydr_.yaml ... DONE
7 #### Cannot stat '/mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-
8 #### Ydr_.yaml': No such file or directory at /usr/lib/perl5/site_perl/
9 #### 5.8.8/File/Rsync/Mirror/Recentfile.pm line 1558.
10 #### unlink0: /mirrors/CPAN/authors/.FRMRecent-RECENT-Z.yaml-Ydr_.yaml is
11 #### gone already at cpan-pause.pl line 0
13 Running without skip-deletes now but cannot reproduce.
15 2009-04-15 Andreas J. Koenig <andk@cpan.org>
17 * consider state directory such we can restart after a ^C or come again
20 * consider the effect of ^C. Is it safe?
22 * want to use Perl::Version to up the versions on every release. Only if
23 things have changed, of course. Does this actually work?
25 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun
26 /usr/local/perl-5.10-g/bin/perl-reversion -dryrun -bump lib/File/Rsync/Mirror/Recentfile.pm
28 Must do the bookkeeping myself when this should happen, maybe based on git?
30 I think I prefer the same version number for all pm files. But one also
31 needs a routine for setting them. Ahh:
33 /usr/local/perl-5.10-g/bin/perl-reversion -current 0.0.4 -set 0.0.5 lib/**/*.pm
39 cpan[2]> m /Mirror::Recent/
40 Module = File::Rsync::Mirror::Recent (ANDK/File-Rsync-Mirror-Recent-0.0.4.tar.bz2)
41 Module = File::Rsync::Mirror::Recentfile (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
42 Module = File::Rsync::Mirror::Recentfile::Done (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
43 Module = File::Rsync::Mirror::Recentfile::FakeBigFloat (ANDK/File-Rsync-Mirror-Recent-0.0.2.tar.bz2)
46 The META.yml was generated by ExtUtils::MakeMaker version 6.5101
48 The 0.0.2 release has a META.yml by ExtUtils::MakeMaker version 6.42 and without "provides".
50 I just tried MM 6.5102 and it again produces only a provides for Recent,
51 not for the other modules.
53 My own fault. FIXED in the Makefile.PL.
55 2009-04-13 Andreas J. Koenig <andk@cpan.org>
57 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
59 * is it true (as stated at
60 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
61 that rsync can lead to errors when upstream changes before client sync
62 has completed? Can we see the error when we run this on upstream:
65 use Time::HiRes qw(time);
67 open my $fh, ">", "changingfile.txt.new" or die;
68 print $fh (time."\n") x 1000000;
70 rename "changingfile.txt.new", "changingfile.txt" or die;
74 And on the receiving end:
77 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
81 I cannot get it to fail. Apparently it is sufficient when upstream
82 always writes atomically (which of course is mandatory)?
86 > https://fedorahosted.org/InstantMirror/
87 > https://www.redhat.com/mailman/listinfo/instantmirror-list
88 > irc.freenode.net channel #instantmirror
90 A cool name for a project. Inspires me to write a few sentences about
91 bittorrent's role in the grand picture.
93 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
94 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
96 2009-04-12 Andreas J. Koenig <andk@cpan.org>
98 * Interesting last minute bug during real download testing: the output
99 isn't as pretty anymore, it seems that more work is being done than
102 Introducing a highlevel _alluptodate method helped a lot in debugging
105 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
106 different on 1h file than on the other files. This lead to frequent
107 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
110 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
112 * Todo: try if we can be faster with a native float package. I'm really
113 glad to have a machine exposing perl floating point bugs, it forced me
114 to retract this item from the changes file: "several speedups in the
115 fakebigfloat code" before I get cpan testers fails.
117 2009-04-11 Andreas J. Koenig <andk@cpan.org>
119 * Bug: when the dirtymark changes upstream, then the downstream server
120 notices it immediately and mirrors the "1h" file but then leaves the
121 rmirror loop and continues with "6h" on the next round through. This
122 behaviour goes on until it reaches the Z file. Because we eat the Z file
123 piecemeal we leave the loop after a while but after that everybody seems
124 to have forgotten that there is some work left to be done. Bad, bad,
127 ETOOCONFUSING better name?
129 Recentfile::get_remote_recentfile_as_tempfile OK
130 Recentfile::resolve_recentfilename split_rfilename DONE
132 Recent::_principal_recentfile_object _principal_recentfile_fromremote
133 Recent::principal_recentfile OK
134 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
136 some void, some not, some not void but called in void context.
138 2009-04-10 Andreas J. Koenig <andk@cpan.org>
140 * Todo: rename rrr-dirtyupdate to rrr-update and add an option --dirty
141 so we can use it for normal update with a single file.
143 * Bug? When the dirtymark changes then the second tier hosts quickly
144 reset their DONE state and restart mirroring. But then they mirror
145 potentially outdated index files with inconsistent dirtymark because the
146 first tier box may be a bit behind on the large recentfiles.
148 This means we need a brake when iterating over the recentfiles that
149 refuses to use a recentfile with the wrong dirtymark.
151 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
153 * install the bin/* files and talk about rrr-overview somewhere.
155 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
156 created outside the target tree.
158 * Barbie asks for better logging capabilities
160 2009-03-29 Andreas J. Koenig <andk@cpan.org>
162 * known bug: cannot configure the Recentfile objects to keep delete
163 events. How to call the accessor for that? keep_delete_objects_forever?
165 Sounds acceptable. FIXED (but untested).
167 * known bug: keeps temporary index files lying around.
169 2009-03-24 Andreas J. Koenig <andk@cpan.org>
171 * Some equivalent for
173 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
177 2009-03-22 Andreas J. Koenig <andk@cpan.org>
179 * k81 is again client of k75 with different parameters.
181 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
182 about it all the time and nobody reacts accordingly; (2) lots of temp
183 files get created and not removed; culprit the new call to
184 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
185 caller has to remove the tempfile after use.
187 Need the drawing board.
189 PARTLY FIXED: retracting the idea to call
190 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
191 and unseed such that uptodate now means "we have mirrored this
192 recentfile's bunch of files" and seeded means "somebody believes the
193 index file for this rf needs be refreshed". This seems to work now.
195 Still collecting temp files that nobody cares to depollute.
197 2009-03-21 Andreas J. Koenig <andk@cpan.org>
199 * Bug?: should it be harder than it is atm to set the timestamp to the
202 * bug with native integers:
206 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
210 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
213 Native integer broke when native math was turned off. FIXED
215 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
216 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
219 Yes, records out of order:
222 447004 epoch: 995885533
223 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
226 447008 epoch: 995890358
227 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
230 447012 epoch: 995892221
231 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
234 FIXED with sanity check and later with the integer fix.
236 * Bug: want the index files in a .recent directory
238 * Bug: lots of dot files are not deleted in time
240 * possible test case: can a delete change the timestamp? This would
241 probably break the order of events.
243 2009-03-20 Andreas J. Koenig <andk@cpan.org>
245 * 1233701831.34486 what's so special about this number/string? It
247 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
248 atm lives in Y,Q, and Z.
250 It is the first entry after id/--skip-locking which has timestamp
251 1234164228.11325 which represents a file that doesn't exist anymore.
255 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
256 _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
257 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
258 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
259 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
260 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
261 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
262 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
263 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
264 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
265 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
266 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
267 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
269 FIXED, it was the "--skip-locking" file where manual intervention was participating
271 * bug on the mirroring slave: when the dirtymark gets increased we
272 probably do not reset the done intervals. The mirrorer stays within
273 tight bounds where it tries to sync with upstream and never seems to
274 finish. In the debugging state file I see lots of identical intervals
275 that do not get collapsed. When I restart the mirrorer it dies with:
277 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
278 _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
279 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
280 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
281 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
282 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
283 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
284 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
285 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
286 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
288 and debugging stands at
294 epoch: 1237400802.5789
295 path: id/J/JO/JOHND/CHECKSUMS
298 epoch: 1237400807.97514
299 path: id/M/MI/MIYAGAWA/CHECKSUMS
302 epoch: 1237400817.94363
303 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
311 and it is reproducable.
313 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
314 pause has a fresh one.
316 FIXED, it was the third anded term in each of the ifs in the IV block in
317 _register_one: with that we make sure that we do not stamp on valuable
320 2009-03-17 Andreas J. Koenig <andk@cpan.org>
322 * done: verified the existence of the floating point bug in bleadperl
323 and verified that switching from YAML::Syck to YAML::XS does not resolve
326 BTW, the switch was doable with
328 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
330 and should be considered as a separate TODO
332 * todo: integrate a dirty update with two aggregate calls before
333 unlocking for frictionless dirtying DONE
335 * todo: start the second rsync daemon on pause DONE
337 * todo: move index files to .recent: this cannot simply be done by
338 setting filenameroot to .recent/RECENT. Other parts of the modules rely
339 on the fact that dirname(recentfile) is the root of the mirrored tree.
341 2009-03-16 Andreas J. Koenig <andk@cpan.org>
343 * What was the resolution of the mirror.pl delete hook bug? Do we call
344 the delete hook when pause removes a file from MUIR?
346 * Today on pause: Updating 2a13fba..29f284d and installing it for
347 /usr/local/perl-5.10.0{,-RC2}
349 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
350 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
352 2009-03-15 Andreas J. Koenig <andk@cpan.org>
354 * currently recent_events has the side effect of setting dirtymark
355 because it forces a read on the file. That should be transparent, so
356 that the dirtymark call always forces a cache-able(?) read.
358 * The bug below is -- after a lot of trying -- not reproducible on a
359 small script, only in the large test script. The closest to the output
365 my $x = "01237123229.8814";
370 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
372 $l = "1237123231.22458";
373 $r = "1237123231.22458";
375 Devel::Peek::Dump $l;
376 Devel::Peek::Dump $r;
377 Devel::Peek::Dump $x = $l <=> $r;
381 The checked in state at c404a85 fails the test with my
382 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
383 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
384 in the test script always consistent.
386 * Todo: write a test that inserts a second dirty file with an already
387 existing timestamp. DONE
389 * Bug in perl 5.10 on my 64bit box:
391 DB<98> Devel::Peek::Dump $l
392 SV = PVMG(0x19e0450) at 0x142a550
394 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
396 NV = 1237123231.22458
397 PV = 0x194ce70 "1237123231.22458"\0
401 DB<99> Devel::Peek::Dump $r
402 SV = PVMG(0x19e0240) at 0x142a3e8
404 FLAGS = (PADMY,POK,pPOK)
407 PV = 0x19ff900 "1237123231.22458"\0
411 DB<100> Devel::Peek::Dump $l <=> $r
412 SV = IV(0x19ea6e8) at 0x19ea6f0
414 FLAGS = (PADTMP,IOK,pIOK)
417 DB<101> Devel::Peek::Dump $l
418 SV = PVMG(0x19e0450) at 0x142a550
420 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
422 NV = 1237123231.22458
423 PV = 0x194ce70 "1237123231.22458"\0
427 DB<102> Devel::Peek::Dump $r
428 SV = PVMG(0x19e0240) at 0x142a3e8
430 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
432 NV = 1237123231.22458
433 PV = 0x19ff900 "1237123231.22458"\0
437 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
439 * Todo: reset "done" or "covered" and "minmax" after a dirty operation? DONE
441 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
443 * $obj->merge ($other) needs to learn about equal epoch which may happen
444 since dirty_epoch intruded.
446 * Wontfix anytime soon: I think we currently do not support mkdir. Only
449 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
451 * Todo: continue working on update(...,$dirty_epoch). It must be
452 followed by a fast_aggregate! DONE
454 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
456 * maybe we need a closest_entry or fitting_interval or something like
457 that. We want to merge an event into the middle of some recentfile.
458 First we do not know which file, then we do not know where to lock,
459 where to enter the new item, when and where to correct the dirtymark.
461 So my thought is we should first find which file.
463 Another part of my brain answers: what would happen if we would enter
464 the new file into the smallest file just like an ordinary new event,
465 just as an old event?
467 (1) we would write a duplicate timestamp? No, this would be easy to
470 (2) we would make the file large quickly? Yes, but so what? We are
471 changing the dirtymark, so are willing to disturb the downstream hosts.
473 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
475 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
479 https://rt.cpan.org/Ticket/Display.html?id=41199
481 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
483 Strangest output being HAKANARDO who managed to upload
485 Here is a better oneliner that includes also the first line of each
488 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
490 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
492 * memory leak in the syncher? It currently weighs 100M.
496 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
499 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
501 * lookup by epoch and by path and use this ability on the pause to never
502 again register a file twice that doesn't need it. Let's call it
505 * after the dirtymark is done: fill up recentfiles with fake (historic)
506 entries; fill up with individual corrections; algorithm maybe to be done
507 with bigfloat so that we can always place something in the middle
508 between two entries. Before we must switch to bigfloat we could try to
509 use Data::Float::nextup to get the.
511 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
514 * dirtymark now lives in Recentfile, needs to be used in rmirror.
516 * find out why the downloader died after a couple of hours without a net
517 connection. Write a test that survives the not-existence of the other
520 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
522 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
523 it makes less sense with tight coupling for secondary files. But asking
524 the server what the current epoch is might be cheaper on HTTP than on
525 rsync. (Needs to be evaluated)
527 * remove the 0.00 from the verbose overview in the Merged column in the
530 * write tests that expose the problems of the last few days: cascading
531 client/server roles, tight coupling for secondary RFs, deletes after
534 * Some day we might want to have policy options for the slave:
535 tight/loose/no coupling with upstream for secondary RFs. tight is what
536 we have now. loose would wait until a gap occurs that can be closed.
538 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
540 * revisit all $rfs->[$i+1] places if they now make sense still
542 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
544 * another bug is the fact that the mirror command deletes files before
545 it unhides the index file, thus confusing downstream slaves. We must not
546 delete before unhiding and must delete after unhiding. FIXED.
548 * new complication about the slave that is playing a server role.
549 Currently we mirror from newest to oldest with a hidden temporary file
550 as index. And when one file is finished, we unhide the index file.
551 Imagine the cascading server/slave is dead for a day. It then starts
552 mirroring again with the freshest thing and unhides the freshest index
553 file when it has worked through it. In that moment it exposes a time
554 hole. Because it now works on the second recentfile which is still
557 We currently do nothing special to converge after such a drop out. At
558 least not intentionally and robustly and thought through.
560 The algorithm we use to seed the next file needs quite a lot of more
561 robustness than it currently has. Something to do with looking at the
562 merged element of the next rf and when it has dropped off, we seed
563 immediately. And if it ramains dropped off, we seed again, of course.
565 Nope, looking from smaller to larger RFS we look at the merged element
566 of this RF and at the minmax/max element of the next RF. If that
567 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
570 And we need a public accessor seed and unseed or seeded. But not the mix
571 of public and private stuff that then is used behind the back.
573 And then the secondary* stuff must go.
575 And we must understand what the impact is on the DONE system. Can it go
576 unnoticed that there was a hole? And could the DONE system have decided
577 the hole is covered? This should be testable with three directories where
578 the middle stops working for a while. Done->merge is suspicious, we must
579 stop it from merging non-conflatable neighbors due to broken continuity.
583 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
585 * Slaven suggests to have the current epoch or the whole current
586 recentfile available from the HTTP server and take it away with
587 keepalive. This direction goes the granularity down to subseconds.
589 We might want to rewrite everything to factor out transport and allow
590 the whole thing to run via HTTP.
592 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
594 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
597 * smoker on k81 fetching from k75 to verify cascading works. See
598 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
600 * maybe the loop should wait for CHECKSUMS file after every upload. And
601 CPAN.pm needs to deal with timestamps in the future.
603 * do not forget the dirtymark!
605 Text: have a new flag on recentfiles with the meaning: if this
606 changes, you're required to run a full rsync over all the files. The
607 reason why we set it would probably be: some foul happened. we injected
608 files in arbitrary places or didn't inject them although they changed.
609 The content of the flag? Timestamp? The relation between the
610 recentfiles would have to be inheritance from the principal, because any
611 out of band changes would soon later propagate to the next recentfile.
613 By upping the flag often one can easily ruin the slaves.
615 last out of band change? dirtymark?
617 Anyway, this implies that we read a potentially existing recentfile
620 And it implies that we have an eventloop that keeps us busy in 2-3
621 cycles, one for current stuff (tight loop) and one for the recentfiles
622 (cascade when principal has changed), one for the old stuff after a
625 And it implies that the out-of-band change in any of the recentfiles
626 must have a lock on the principal file and there is the place to set the
629 * start a FAQ, especially quick start guide questions. Also to aid those
630 problematic areas where we have no good solution, like the "links"
633 * wish feedback when we are slow.
637 * Remove a few DEBUG statements.
639 * The multiple-rrr way of doing things needs a new option to rmirror,
640 like piecemeal or so. Not urgent because after the first pass through,
641 things run smoothely. It's only ugly during the first pass.
643 * I have the suspicion that the code is broken that decides if the
644 neighboring RF needs to be seeded. I fear when too much time has gone
645 between two calls (in our case more than one hour), it would not seed
646 the neighbor. Of course this will never be noticed, so we need a good
649 * local/localroot confusion: I currently pass both options but one must
652 * accounts for early birds on PAUSE rsync daemon.
654 * hardcoded 20 seconds
656 * who mirrors the index? DOING now.
658 * which CPAN mirrors offer rsync?
660 * visit all XXX, visit all _float places
662 * rename the pathdb stuff, it's too confusing. No idea how.
664 * rrr-inotify, backpan, rrr-register
666 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
668 * current bugs: the pathdb seems to get no reset, the seeding of the
669 secondaryttl stuff seems not to have an effect. Have helped myself with
670 a rand(10), need to fix this back. So not checked in. Does the rand
673 The rand thing helps. The secondaryttl stuff was in the wrong line,
676 The pathdb stuff was because I called either _pathdb or __pathdb on the
677 wrong object. FIXED now.
679 * It's not so beautiful if we never fetch the recentfiles that are not
680 the principal, even if this is correct behaviour. We really do not need
681 them after we have fetched the whole content.
683 OK, we want a switch for that: secondaryttl DONE
685 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
687 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
688 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
689 it lets 40 follow. FIXED
691 * See that the long running process really only updates the principal
692 file unless it has missed a timespan during which something happened. If
693 nothing happened, it must notice even when it misses the timespan. DONE
695 * we must throw away the pathdb when we have reached the end of Z. From
696 that moment we can have a very small pathdb because the only reason for
697 a pathdb is that we know to ignore old records in old files. We won't
698 need this pathdb again before the next full pass over the data is
699 necessary and then we will rebuild it as we go along. DONE
701 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
703 * I think, Done::register_one is doing wrong in that it does not
704 conflate neighboring pieces. The covered() method cannot do this because
705 it has no recent_events array at hand. But register_one has it and could
706 do it and for some reason misses to do it (sometimes).
708 This means that the three tests I just wrote can probably not survive
709 because they test with an already broken Done structure.
711 The art now is to detect how it happens, then to reproduce, then write a
714 So from the logfile this is what happens: we have a good interval with
715 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
716 top of it. Locally we now mirror F2 and open a new done interval for it.
717 Then we mirror F1 but this time with the timestamp T1b. And when we then
718 try to close the gap, we do not find T1 but instead something older. We
719 should gladly accept this older piece and this would fix this bug.
723 * bug to fix: when the 1h file changes while rmirror is running, we do
724 correctly sync the new files but never switch to the 6h file but rather
725 stay in a rather quick loop that fetches the 1h file again and again.
727 Is it possible that we initialize a new object? Or does
728 get_remote_recentfile_as_tempfile overwrite something in myself?
730 Want a new option: _runstatusfile => $file which frequently dumps the
731 state of all recentfiles to a file.
735 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
737 * Todo: now teach update to verify the timestamp is about to write
738 against the previous and use _increase_a_bit if it doesn't comply with
739 strict monotony. DONE
741 * The problem of rounding. So far perl's default precision was
742 sufficient. One day it won't be. FakeFloat has an easy job when it is
743 only reading and other machines have written correctly. But when we want
744 to write a floating point number that is a bit larger than the other
745 one, then we need our own idea of precision.
747 Slaven said: just append a "1". This might be going towards the end of
748 usability too quickly. I'd like something that actually uses the decimal
749 system. Well, appending a 1 also does this but...
751 E.g. we have 1.0. nextup on this architecture is starting with
752 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
753 taken the 1.0000000000000003 and the next user comes and the time tells
754 him 1.0 again. He has to beat my number without stepping over the
755 nextup. This is much less space than I had when I chose 1,2,3.
757 What is also irritating is that nextup is architecture dependent. The
758 128 bit guy must choose very long numbers to fit in between whereas the
759 other one with 16 bit uses larger steps. But then the algorithm is the
760 same for both, so that would be a nice thing.
762 I see two situation where we need this. One is when Time::HiRes returns
763 us a value that is <= the last entry in our recentfile. In this case
764 (let's call it the end-case) we must fill the region between that number
765 and the next higher native floating point number. The other is when we
766 inject an old file into an old recentfile (we would then also set a new
767 dirtymark). We find the integer value already taken and need a slightly
768 different one (let's call it the middle-case). The difference between
769 the two situations is that the next user will want to find something
770 higher than my number in the end-case and something lower than my number
773 So I suggest we give the function both a value and an upper bound and it
774 calculates us a primitive middle. The upper bound in the middle-case is
775 the next integer. The upper bound on the end-case is the nextup floating
776 point number. But the latter poses another problem: if we have occupied
777 the middle m between x and nextup(x), then the nextup(m) will probably
778 not be the same as nextup(x) because some rounding will take place
779 before the nextup is calculated and when the rounding reaches the
780 nextup(x), we will end up at nextup(nextup(x)).
782 So we really need to consider the nextup and the nextdown from there and
783 then the middle and that's the number we may approach asymptotically.
786 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
788 * consider deprecating the use of RECENT.recent as a symlink. It turns
789 out to need extra hoops with the rsync options and just isn't worth it.
790 Or maybe these extra hoops are needed anyway for the rest of the tree?
791 Nope, can't be the case because not all filesystems support symlinks.
793 But before doing the large step, I'll deprecate the call of
794 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
795 only introduced to resolve RECENT.recent and complicates the routine far
796 beyond what it deserves.
798 DONE. Won't deprecate RECENT.recent, just moved its handling to the
801 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
803 * I think it's a bug that the rsync_option links must be set to true in
804 order to support RECENT.recent and that nobody cares to set it
805 automatically. Similar for ignore_link_stat_errors.
807 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
809 * Todo: find all todos together and make a plan what is missing for a
812 - verifytree or something like that. fsck maybe.
814 - rersyncrecent, the script itself? What it do?
816 - a way to only mirror the recentfiles without mirroring the whole
817 remote system such that people can decide to mirror only partially see
818 also 2008-08-30. .shadow-xxx directory? this also needed for a
819 filesystem that is still incomplete and might need the mirrorfiles for
822 - long living objects that mirror again and again. Inject something
823 into ta, see how it goes over to tb.
825 - how do we continue filling up the DONE system when we use an object
826 for the second time? "fully covered" and "uptodate" or new terminology.
828 - overview called on the wrong file should be understandable
830 - the meta data field that must change when we fake something up so that
831 the downstream people know they have to re-fetch everything.
833 - how tolerant are we against missing files upstream? how do we keep
834 track? there are legitimate cases where we did read upstream index right
835 before a file got deleted there and then find that file as new and want
836 it. There are other cases that are not self healing and must be tracked
839 - how, exactly, do we have to deal with deletes? With rsync errors?
841 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
842 authors) failed: No such file or directory (2)
844 The file above is a delete in 1h and a new in file 1M and the
845 delete in the locally running rmirror did not get propagated to the 1M
846 object. Bug. And the consequence is a standstill.
848 It seems that a slave that works with a file below the principal needs
849 to merge things all the way up to get rid of later deletes. Or keep
850 track of all deletes and skip them later. So we need a trackdeletes.pm
851 similar to the done.pm?
853 see also 2008-08-20 about spurious deletes that really have no add
854 counterpart and yet they are not wrong.
856 - consider the effect when resyncing the recentfile takes longer than
857 the time per loop. Then we never rsync any file. We need to diagnose
858 that and force an increase of that loop time. But when we later are fast
859 enough again because the net has recovered, then we need to switch back
860 to original parameters. ERm, no, it's enough to keep syncing at least
861 one file before refetching an index file.
863 - remember to verify that no temp files are left lying around and the
866 - status file for not long running jobs that want to track upstream with
869 - revisit all XXX _float areas and study Sub::Exporter DONE
871 - persistent DB even though we just said we do not need it. Just for
872 extended capabilities and time savings when, for example, upstream
873 announces a reset and we get new recentfiles and could then limit
874 ourselves to a subset of files (those that have a changed epoch) in a
875 first pass and would only then do the loop to verify the rest. Or
878 * Todo: aggregate files should know their feed and finding the principal
879 should be done stepwise. (?)
881 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
882 left locks around. DONE
884 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
886 * maybe extend the _overview so that it always says if and where the
887 last file is in the next file and where the next event in the next rf
888 would lie. No, don't like this anymore. REJECT
890 * take the two new redundant tests out again, only the third must
893 * Todo: add a sanity check if the merged structure is really pointing to
894 a different rf and that this different rf is larger. DONE
896 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
898 * now test, if they are overlapping. And test if there is a file in the
899 next rf that would fit into this rf's interval.
901 1h 1222324012.8474 1222322541.7963 0.4086
902 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
903 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
904 1W 1222313218.3626 1221708477.5829 167.9835
906 I suspect that somebody writes a merged timestamp without having merged
907 and then somebody else relies on it.
909 If aggregate is running, the intervals must not be extravagated, if it
910 is not running, there must not be bounds, the total number of events in
911 the system must be counted and must be controlled throughout the tests.
912 That the test required the additional update was probably nonsense,
913 because aggregate can cut pieces too. FIXED & DONE
915 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
917 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
920 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
922 * Most apparent bug at the moment is that the recentfiles are fetched
923 too often. Only the principal should be fetched and if it has not
924 changed, the others should not be refetched. ATM I must admit that I'm
925 happy that we refetch more often than needed because I can more easily
926 fix bugs while the thing is running.
928 * Let's say, 1220474966.19501 is a timestamp of a file that is already
929 done but the done system does not know about it. The reason for the
930 failure is not known and we never reach the status uptodate because of
931 this. We must get over it.
933 Later it turns out that the origin server had a bug somewhere.
934 1220474966.19042 came after 1220474966.19501. Or better: it was in the
935 array of the recentfile one position above. The bug was my own.
937 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
939 * There is the race condition where the server does a delete and the
940 slave does not yet know and then tries to download it because he sees
941 the new. So for this time window we must be more tolerant against
942 failure. If we cannot download a file, we should just skip it and should
943 not retry immediately. The whole system should discover the lost thing
944 later. Keeping track with the DONE system should really be a no brainer.
946 But there is something more: the whole filesystem is a database and the
947 recentfiles are one possible representation of it. It's a pretty useful
948 representation I think that's why I have implemented something around
949 it. But for strictly local operation it has little value. For local
950 operation we would much rather have a database. So we would enter every
951 recentfile reading and every rsync operation and for every file the last
952 state change and what it leads to. Then we would always ignore older
953 records without the efforts involved with recentfiles.
955 The database would have: path,recentepoch,rsyncedon,deletedon
957 Oh well, not yet clear where this leads to.
959 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
961 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
962 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
964 * statusfile, maybe only on demand, alone to have a sharp debugging
965 tool. It is locked and all recentfiles dump themselves into it and we
966 can build a viewer that lets us know where we stand and what's inside.
968 * remember: only the principal recentfile needs expiration, all others
969 shall be expired by principal if it discovers that something has move
972 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
974 * Always check if we stringify to a higher value than in the entry
977 * And in covered make an additional check if we would be able to see a
978 numerical difference between the two numbers and if we can't then switch
979 to a different, more expensive algorithm. Do not want to be caught by
980 floating surprises. DONE
982 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
984 * caching has several aspects here: we can cache the interval of the
985 recentfile which only will change when the mtime of the file changes. We
986 must re-mirror the recentfile when its ttl has expired. Does have_read
987 tell you anything? It counts nothing at all. Only the mtime is
988 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
989 thing the have_mirrored because it is unrelated to the mtime.
991 * Robustness of floating point calculations! I always thought that the
992 string calculated by the origin server for the floating representation
993 of the epoch time is just a string. When we convert it to a number and
994 later back to a string, the other computer might come to a different
995 conclusion. This must not happen, we want to preserve it under any
996 circumstances. I will have to write tests with overlong sequences that
997 get lost in arithmetic and must see if all still works well. DONE
999 But one fragile point remains: if one host considers a>b and the other
1000 one considers them == but no eq. To prevent this, we must probably do
1001 some extra homework. DONE
1003 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1005 * the concept of tracking DONE needs an object per recentfile that has
1006 something like these methods:
1008 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
1010 covered() register() covered()
1012 The unclear thing is how we translate points in time into intervals. We
1013 could pass a reference to the current recent_events array when running
1014 we_have(xxx) and let the DONE object iterate over it such that it only
1015 has to store a list of intervals that can melt into each other. Ah, even
1016 passing the list together with a list of indexes seems feasiable.
1018 Or maybe ask for the inverted list?
1020 Whenever the complete array is covered by the interval we say we are
1021 fully covered and if the recentfile is not expired, we are uptodate.
1023 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1025 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1027 * need a way to "return" the next entry after the end of a list. When
1028 the caller says "before" or "after" we would like to know if he could
1029 cover that interval/threshold or not because this influences the effect
1030 of a newer timestamp of that recentfile. DONE with $opt{info}.
1032 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1034 * one of the next things to tackle: the equivalent of csync2 -TIXU.
1036 loop implies tixu (?). Nope, something like --statefile decides. Per
1039 T test, I init, X including removals, U nodirtymark
1041 So we have no concept of dirtymarks, we only trust that since we are
1042 running we have observed everything steadily. But people will not let
1043 this program run forever so we must consider both startup penalty and
1044 book keeping for later runs. We keep this for later. For now we write a
1045 long running mirror that merges several intervals.
1047 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1049 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
1052 * rersyncrecent, the script: default to one week. The name of the switch
1053 is --after. Other switches? --loop!
1055 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1057 * need a switch --skip-deletes (?)
1059 * need a switch --enduser that tells us that the whole tempfile
1060 discipline is not needed when there is no downstream user. (?)
1062 Without this switch we cannot have a reasonable recent.pl that just
1063 displays the recent additions. Either we accept to download everything.
1064 Or we download temporary files without the typical rsync protocol
1067 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1068 File::Temp::tempdir, this might be a win.
1070 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1072 * apropos missing: we have no push, we never know the downstream
1073 servers. People who know their downstream hosts and want to ascertain
1074 something will want additional methods we have never thought about, like
1075 update or delete a certain file.
1077 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1079 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1080 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1083 * rsync options: it's a bit of a pain that we usually need several rsync
1084 options, like compress, links, times, checksum and that there is no
1085 reasonable default except the original rsync default. I think wee can
1086 safely assume that the rsync options are shared between all recentfile
1087 instances within one recent tree.
1089 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1091 * deletes: if a delete follows an add quickly enough it may happen that
1092 a downstream mirror did not see the add at all! It seems this needs to
1093 be mentioned somewhere. The point here is that even if the downstream is
1094 never missing the principal timeframe it may encounter a "delete" that
1095 has no complimentary "add" anywhere.
1097 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1099 * I suspect the treat of metadata is incorrect during read or something.
1100 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1101 contained more than 6 hours worth of data. At 06:08 we merged into the
1102 1d file. We need to take snapshots of the 6h file over the course of an
1103 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1106 Much worse: watching the 1h file: right at the moment (at 06:35) it
1107 covers 1218867584-1219120397 which is 70 hours.
1109 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1110 08:19:44 2008, that is when I checked out last time, so it seems to be
1111 aggregating and never truncating?
1113 No, correct is: it is never truncating; but wrong is: it is aggregating.
1114 It does receive a lot of events from time to time from a larger file.
1115 Somehow a large file gets merged into the small one and because the
1116 "meta/merged" attribute is missing, nobody is paying attention. I
1117 believe that I can fix this by making sure that metadata are honoured
1118 during read. DONE and test adjusted.
1120 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1122 * grand renaming plan
1124 remotebase => remoteroot to fit well with localroot DONE
1125 local_path() => localroot seems to me should already work DONE
1126 recentfile_basename => rfilename no need to stress it has no slash DONE
1128 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1129 kern? It would anyway need a deprecation cycle because it is an
1130 important constructor.
1132 * I like the portability that Data::Serializer brings us but the price
1133 is that some day we might find out that it is slowing us a bit. We'll
1136 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1138 * should we not enter the interval of the principal (or the interval of
1139 the merging file?) in every aggregated/merged file?
1141 * we should aim at a first release and give up on thinking about
1142 sanitizing stuff and zloop. Let's just admit that a full traditional
1143 rsync is the only available sanitizer ATM. Otherwise it's complicated
1144 stuff: sanitizing on the origin server, sanitizing on the slaves,
1145 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1146 get the basics out before this becomes a major cause for mess.
1148 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1150 * On OSes not supporting symlinks we expect that RECENT.recent contains
1151 the contents of the principal recentfile. Actually this is identical on
1152 systems supporting symlinks. Simple, what follows from that is that we
1153 need to keep the serializer in the metadata because we cannot read it
1154 from the filename, doesn't it? Of course not. It's a chicken and egg
1155 problem. This leaves us with the problem to actually parse the
1156 serialized data to find out in which format it is. So who can do the 4
1157 or 5 magics we wanted to support? File::LibMagic?
1159 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1161 * remotebase and recentfile_basename are ugly names. Now that we need a
1162 word for the shortest/principal/driving recentfile too we should do
1165 localroot is good. rfile is good. local_path() is bad, local_path($path)
1166 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1169 Up to now remotebase was the string that described the remote root
1170 directory in rsync notation, like pause.perl.org::authors. And
1171 recentfile_basename was "RECENT-1h.yaml".
1173 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1175 * The test that was added in today's checkin is a good start for a test
1176 of rmirror. We should have more methods in Recent.pm: verify,
1177 addmissingfiles. We should verify the current tree, then rmirror it and
1178 then verifytree the copy. We could then add some arbitrary file and let
1179 it be discovered by addmissingfiles, then rmirror again and then
1180 verifytree the copy again.
1182 Then we could start stealing from csync2 sqlite database [no port to
1183 OSX!] and fill a local DB. And methods to compare the database with the
1184 recentfiles. Our strength is that in principle we could maintain state
1185 with a single float. We have synced up to 1234567890.123456. If the Z
1186 file does not add new files all we have to do is mirror the new ones and
1189 This makes it clear that we should extend current protocol and declare
1190 that we cheat when we add files too late, just to help the other end
1191 keeping track. Ah yes, that's what was meant when zloop was mentioned
1194 Maybe need to revisit File::Mirror to help me with this task.
1196 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1198 * There must be an allow-me-to-truncate flag in every recentfile.
1199 Without it one could construct a sequence of updates winning the locking
1200 battle against the aggregator. Only if an aggregator has managed to
1201 merge data over to the next level, truncating can be allowed. DONE with
1204 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1206 * We should probably guarantee that no duplicates enter the aggregator
1209 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1211 * To get merge operation faster would need a good benchmark test. What
1212 02 spits out isn't reliable enough and is dominated by many other
1215 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1216 Author: Andreas J. Koenig <k@k75.(none)>
1217 Date: Sat Aug 2 07:58:04 2008 +0200
1221 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1222 Author: Andreas J. Koenig <k@k75.(none)>
1223 Date: Sat Aug 2 11:40:29 2008 +0200
1225 there was a lot of trying but the effect is hardly measurable with
1228 * overhead of connecting seems high. When setting
1229 max_files_per_connection to 1 we see that.
1231 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1233 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1235 25138.2215600014/3600 = 6.98283932222261
1237 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1238 be a bug in the aggregator. FIXED
1240 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1242 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1243 we have it, can we mirror it, mirror it!
1245 I fear this needs a new class which might be called
1246 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1247 and treat them as an entity. I realize that a single recentfile may be
1248 sufficient for certain tasks and that it is handy for the low level
1249 programmer but it is not nice to use. If there is a delete in the 1h
1250 file then the 6h file still contains it. Seekers of the best information
1251 need to combine at least some of the recentfiles most of the time.
1253 There is the place for the Z loop!
1255 But the combination is something to collect in a database, isn't it. Did
1256 csync2 just harrumph?
1258 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1260 * it just occurred to me that hosts in the same mirroring pool could
1261 help out each other even without rewriting the recentfile. Just fetch
1262 the stuff to mirror from several places, bingo. But that's something
1263 that should rather live in a separate package or in rsync directly.
1265 * cronjobs are unsuited because with ntp they would all come at the full
1266 minute and disturb each other. Besides that I'd hate to have a backbone
1267 with more than a few seconds latency.
1269 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1271 * a second rsync server with access control for PAUSE. Port? 873 is the
1272 standard port, let's take 8873.
1274 * if there were a filesystem based on this, it would have a slow access
1275 to inexistent files. It would probably provide wrong readdir (only based
1276 on current content) or also a slow one (based on a recentfile written
1277 after the call). But it would provide fast access to existing files. Or
1278 one would deliberately allow slightly blurred answers based on some
1279 sqlite reflection of the recentfiles.
1281 * todo: write a variant of mirror() that combines two or more
1282 recentfiles and treats them like one
1284 * todo: signal handler to remove the tempfile
1286 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1288 * now that we have the symlink I forgot how it should be used in
1291 * the z loop: add missing files to Z file. Just append them (instead of
1292 prepending). So one guy prepends something from the Y file from time to
1293 time and another guy appends something rather frequently. Collecting
1294 pond. When Y merges into Z, things get epoch and the collecting pond
1295 gets smaller. What exactly are "missing files"?
1297 take note of current epoch of the alpha file, let's call it the
1300 find all files on disk
1302 remove all files registered in the recentworld up to recent-ts
1304 remove all files that have been deleted after recent-ts according to
1307 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1309 * rersyncrecent might be a cronjob with a (locked) state file which
1310 contains things like after and maybe last z sync or such?
1312 rrr-mirror might be an alternative name but how would we justify the
1313 three Rs when there is no Re-Rsync-Recent?
1315 With the --loop parameter it is an endless loop, without it is no loop.
1316 At least this is simple.
1318 * todo: new accssor z-interval specifies how often the Z file is updated
1319 against the filesystem. We probably want no epoch stamp on these
1320 entries. And we want to be able to filter the entries (e.g. no
1321 by-modules and by-category tree)
1323 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1325 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1326 into Z. We do not need the other files filled up with filesystem
1329 * need interface to query for a file in order to NOT call update on
1330 PAUSE a second time within a short time.
1332 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1334 * recommended update interval? Makes no sense, is different for
1341 change-log-default-name: "Todo"