1 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * Todo: try if we can be faster with a native float package. I'm really
4 glad to have a machine exposing perl floating point bugs:-(
6 2009-04-11 Andreas J. Koenig <andk@cpan.org>
8 * Bug: when the dirtymark changes upstream, then the downstream server
9 notices it immediately and mirrors the "1h" file but then leaves the
10 rmirror loop and continues with "6h" on the next round through. This
11 behaviour goes on until it reaches the Z file. Because we eat the Z file
12 piecemeal we leave the loop after a while but after that everybody seems
13 to have forgotten that there is some work left to be done. Bad, bad,
16 ETOOCONFUSING better name?
18 Recentfile::get_remote_recentfile_as_tempfile OK
19 Recentfile::resolve_recentfilename split_rfilename DONE
21 Recent::_principal_recentfile_object _principal_recentfile_fromremote
22 Recent::principal_recentfile OK
23 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
25 some void, some not, some not void but called in void context.
27 2009-04-10 Andreas J. Koenig <andk@cpan.org>
29 * Todo: rename rrr-dirtyupdate to rrr-update and add an option --dirty
30 so we can use it for normal update with a single file.
32 * Bug? When the dirtymark changes then the second tier hosts quickly
33 reset their DONE state and restart mirroring. But then they mirror
34 potentially outdated index files with inconsistent dirtymark because the
35 first tier box may be a bit behind on the large recentfiles.
37 This means we need a brake when iterating over the recentfiles that
38 refuses to use a recentfile with the wrong dirtymark.
40 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
42 * install the bin/* files and talk about rrr-overview somewhere.
44 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
45 created outside the target tree.
47 * Barbie asks for better logging capabilities
49 2009-03-29 Andreas J. Koenig <andk@cpan.org>
51 * known bug: cannot configure the Recentfile objects to keep delete
52 events. How to call the accessor for that? keep_delete_objects_forever?
54 Sounds acceptable. FIXED (but untested).
56 * known bug: keeps temporary index files lying around.
58 2009-03-24 Andreas J. Koenig <andk@cpan.org>
62 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
66 2009-03-22 Andreas J. Koenig <andk@cpan.org>
68 * k81 is again client of k75 with different parameters.
70 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
71 about it all the time and nobody reacts accordingly; (2) lots of temp
72 files get created and not removed; culprit the new call to
73 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
74 caller has to remove the tempfile after use.
76 Need the drawing board.
78 PARTLY FIXED: retracting the idea to call
79 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
80 and unseed such that uptodate now means "we have mirrored this
81 recentfile's bunch of files" and seeded means "somebody believes the
82 index file for this rf needs be refreshed". This seems to work now.
84 Still collecting temp files that nobody cares to depollute.
86 2009-03-21 Andreas J. Koenig <andk@cpan.org>
88 * Bug?: should it be harder than it is atm to set the timestamp to the
91 * bug with native integers:
95 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
99 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
102 Native integer broke when native math was turned off. FIXED
104 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
105 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
108 Yes, records out of order:
111 447004 epoch: 995885533
112 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
115 447008 epoch: 995890358
116 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
119 447012 epoch: 995892221
120 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
123 FIXED with sanity check and later with the integer fix.
125 * Bug: want the index files in a .recent directory
127 * Bug: lots of dot files are not deleted in time
129 * possible test case: can a delete change the timestamp? This would
130 probably break the order of events.
132 2009-03-20 Andreas J. Koenig <andk@cpan.org>
134 * 1233701831.34486 what's so special about this number/string? It
136 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
137 atm lives in Y,Q, and Z.
139 It is the first entry after id/--skip-locking which has timestamp
140 1234164228.11325 which represents a file that doesn't exist anymore.
144 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
145 _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
146 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
147 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
148 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
149 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
150 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
151 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
152 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
153 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
154 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
155 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
156 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
158 FIXED, it was the "--skip-locking" file where manual intervention was participating
160 * bug on the mirroring slave: when the dirtymark gets increased we
161 probably do not reset the done intervals. The mirrorer stays within
162 tight bounds where it tries to sync with upstream and never seems to
163 finish. In the debugging state file I see lots of identical intervals
164 that do not get collapsed. When I restart the mirrorer it dies with:
166 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
167 _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
168 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
169 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
170 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
171 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
172 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
173 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
174 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
175 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
177 and debugging stands at
183 epoch: 1237400802.5789
184 path: id/J/JO/JOHND/CHECKSUMS
187 epoch: 1237400807.97514
188 path: id/M/MI/MIYAGAWA/CHECKSUMS
191 epoch: 1237400817.94363
192 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
200 and it is reproducable.
202 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
203 pause has a fresh one.
205 FIXED, it was the third anded term in each of the ifs in the IV block in
206 _register_one: with that we make sure that we do not stamp on valuable
209 2009-03-17 Andreas J. Koenig <andk@cpan.org>
211 * done: verified the existence of the floating point bug in bleadperl
212 and verified that switching from YAML::Syck to YAML::XS does not resolve
215 BTW, the switch was doable with
217 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
219 and should be considered as a separate TODO
221 * todo: integrate a dirty update with two aggregate calls before
222 unlocking for frictionless dirtying
224 * todo: start the second rsync daemon on pause
226 * todo: move index files to .recent: this cannot simply be done by
227 setting filenameroot to .recent/RECENT. Other parts of the modules rely
228 on the fact that dirname(recentfile) is the root of the mirrored tree.
230 2009-03-16 Andreas J. Koenig <andk@cpan.org>
232 * What was the resolution of the mirror.pl delete hook bug? Do we call
233 the delete hook when pause removes a file from MUIR?
235 * Today on pause: Updating 2a13fba..29f284d and installing it for
236 /usr/local/perl-5.10.0{,-RC2}
238 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
239 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
241 2009-03-15 Andreas J. Koenig <andk@cpan.org>
243 * currently recent_events has the side effect of setting dirtymark
244 because it forces a read on the file. That should be transparent, so
245 that the dirtymark call always forces a cache-able(?) read.
247 * The bug below is -- after a lot of trying -- not reproducible on a
248 small script, only in the large test script. The closest to the output
254 my $x = "01237123229.8814";
259 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
261 $l = "1237123231.22458";
262 $r = "1237123231.22458";
264 Devel::Peek::Dump $l;
265 Devel::Peek::Dump $r;
266 Devel::Peek::Dump $x = $l <=> $r;
270 The checked in state at c404a85 fails the test with my
271 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
272 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
273 in the test script always consistent.
275 * Todo: write a test that inserts a second dirty file with an already
276 existing timestamp. DONE
278 * Bug in perl 5.10 on my 64bit box:
280 DB<98> Devel::Peek::Dump $l
281 SV = PVMG(0x19e0450) at 0x142a550
283 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
285 NV = 1237123231.22458
286 PV = 0x194ce70 "1237123231.22458"\0
290 DB<99> Devel::Peek::Dump $r
291 SV = PVMG(0x19e0240) at 0x142a3e8
293 FLAGS = (PADMY,POK,pPOK)
296 PV = 0x19ff900 "1237123231.22458"\0
300 DB<100> Devel::Peek::Dump $l <=> $r
301 SV = IV(0x19ea6e8) at 0x19ea6f0
303 FLAGS = (PADTMP,IOK,pIOK)
306 DB<101> Devel::Peek::Dump $l
307 SV = PVMG(0x19e0450) at 0x142a550
309 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
311 NV = 1237123231.22458
312 PV = 0x194ce70 "1237123231.22458"\0
316 DB<102> Devel::Peek::Dump $r
317 SV = PVMG(0x19e0240) at 0x142a3e8
319 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
321 NV = 1237123231.22458
322 PV = 0x19ff900 "1237123231.22458"\0
326 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
328 * Todo: reset "done" or "covered" and "minmax" after a dirty operation?
330 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
332 * $obj->merge ($other) needs to learn about equal epoch which may happen
333 since dirty_epoch intruded.
335 * Wontfix anytime soon: I think we currently do not support mkdir. Only
338 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
340 * Todo: continue working on update(...,$dirty_epoch). It must be
341 followed by a fast_aggregate!
343 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
345 * maybe we need a closest_entry or fitting_interval or something like
346 that. We want to merge an event into the middle of some recentfile.
347 First we do not know which file, then we do not know where to lock,
348 where to enter the new item, when and where to correct the dirtymark.
350 So my thought is we should first find which file.
352 Another part of my brain answers: what would happen if we would enter
353 the new file into the smallest file just like an ordinary new event,
354 just as an old event?
356 (1) we would write a duplicate timestamp? No, this would be easy to
359 (2) we would make the file large quickly? Yes, but so what? We are
360 changing the dirtymark, so are willing to disturb the downstream hosts.
362 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
364 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
368 https://rt.cpan.org/Ticket/Display.html?id=41199
370 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
372 Strangest output being HAKANARDO who managed to upload
374 Here is a better oneliner that includes also the first line of each
377 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
379 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
381 * memory leak in the syncher? It currently weighs 100M.
385 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
388 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
390 * lookup by epoch and by path and use this ability on the pause to never
391 again register a file twice that doesn't need it. Let's call it
394 * after the dirtymark is done: fill up recentfiles with fake (historic)
395 entries; fill up with individual corrections; algorithm maybe to be done
396 with bigfloat so that we can always place something in the middle
397 between two entries. Before we must switch to bigfloat we could try to
398 use Data::Float::nextup to get the.
400 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
403 * dirtymark now lives in Recentfile, needs to be used in rmirror.
405 * find out why the downloader died after a couple of hours without a net
406 connection. Write a test that survives the not-existence of the other
409 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
411 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
412 it makes less sense with tight coupling for secondary files. But asking
413 the server what the current epoch is might be cheaper on HTTP than on
414 rsync. (Needs to be evaluated)
416 * remove the 0.00 from the verbose overview in the Merged column in the
419 * write tests that expose the problems of the last few days: cascading
420 client/server roles, tight coupling for secondary RFs, deletes after
423 * Some day we might want to have policy options for the slave:
424 tight/loose/no coupling with upstream for secondary RFs. tight is what
425 we have now. loose would wait until a gap occurs that can be closed.
427 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
429 * revisit all $rfs->[$i+1] places if they now make sense still
431 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
433 * another bug is the fact that the mirror command deletes files before
434 it unhides the index file, thus confusing downstream slaves. We must not
435 delete before unhiding and must delete after unhiding. FIXED.
437 * new complication about the slave that is playing a server role.
438 Currently we mirror from newest to oldest with a hidden temporary file
439 as index. And when one file is finished, we unhide the index file.
440 Imagine the cascading server/slave is dead for a day. It then starts
441 mirroring again with the freshest thing and unhides the freshest index
442 file when it has worked through it. In that moment it exposes a time
443 hole. Because it now works on the second recentfile which is still
446 We currently do nothing special to converge after such a drop out. At
447 least not intentionally and robustly and thought through.
449 The algorithm we use to seed the next file needs quite a lot of more
450 robustness than it currently has. Something to do with looking at the
451 merged element of the next rf and when it has dropped off, we seed
452 immediately. And if it ramains dropped off, we seed again, of course.
454 Nope, looking from smaller to larger RFS we look at the merged element
455 of this RF and at the minmax/max element of the next RF. If that
456 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
459 And we need a public accessor seed and unseed or seeded. But not the mix
460 of public and private stuff that then is used behind the back.
462 And then the secondary* stuff must go.
464 And we must understand what the impact is on the DONE system. Can it go
465 unnoticed that there was a hole? And could the DONE system have decided
466 the hole is covered? This should be testable with three directories where
467 the middle stops working for a while. Done->merge is suspicious, we must
468 stop it from merging non-conflatable neighbors due to broken continuity.
472 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
474 * Slaven suggests to have the current epoch or the whole current
475 recentfile available from the HTTP server and take it away with
476 keepalive. This direction goes the granularity down to subseconds.
478 We might want to rewrite everything to factor out transport and allow
479 the whole thing to run via HTTP.
481 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
483 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
486 * smoker on k81 fetching from k75 to verify cascading works. See
487 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
489 * maybe the loop should wait for CHECKSUMS file after every upload. And
490 CPAN.pm needs to deal with timestamps in the future.
492 * do not forget the dirtymark!
494 Text: have a new flag on recentfiles with the meaning: if this
495 changes, you're required to run a full rsync over all the files. The
496 reason why we set it would probably be: some foul happened. we injected
497 files in arbitrary places or didn't inject them although they changed.
498 The content of the flag? Timestamp? The relation between the
499 recentfiles would have to be inheritance from the principal, because any
500 out of band changes would soon later propagate to the next recentfile.
502 By upping the flag often one can easily ruin the slaves.
504 last out of band change? dirtymark?
506 Anyway, this implies that we read a potentially existing recentfile
509 And it implies that we have an eventloop that keeps us busy in 2-3
510 cycles, one for current stuff (tight loop) and one for the recentfiles
511 (cascade when principal has changed), one for the old stuff after a
514 And it implies that the out-of-band change in any of the recentfiles
515 must have a lock on the principal file and there is the place to set the
518 * start a FAQ, especially quick start guide questions. Also to aid those
519 problematic areas where we have no good solution, like the "links"
522 * wish feedback when we are slow.
526 * Remove a few DEBUG statements.
528 * The multiple-rrr way of doing things needs a new option to rmirror,
529 like piecemeal or so. Not urgent because after the first pass through,
530 things run smoothely. It's only ugly during the first pass.
532 * I have the suspicion that the code is broken that decides if the
533 neighboring RF needs to be seeded. I fear when too much time has gone
534 between two calls (in our case more than one hour), it would not seed
535 the neighbor. Of course this will never be noticed, so we need a good
538 * local/localroot confusion: I currently pass both options but one must
541 * accounts for early birds on PAUSE rsync daemon.
543 * hardcoded 20 seconds
545 * who mirrors the index? DOING now.
547 * which CPAN mirrors offer rsync?
549 * visit all XXX, visit all _float places
551 * rename the pathdb stuff, it's too confusing. No idea how.
553 * rrr-inotify, backpan, rrr-register
555 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
557 * current bugs: the pathdb seems to get no reset, the seeding of the
558 secondaryttl stuff seems not to have an effect. Have helped myself with
559 a rand(10), need to fix this back. So not checked in. Does the rand
562 The rand thing helps. The secondaryttl stuff was in the wrong line,
565 The pathdb stuff was because I called either _pathdb or __pathdb on the
566 wrong object. FIXED now.
568 * It's not so beautiful if we never fetch the recentfiles that are not
569 the principal, even if this is correct behaviour. We really do not need
570 them after we have fetched the whole content.
572 OK, we want a switch for that: secondaryttl DONE
574 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
576 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
577 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
578 it lets 40 follow. FIXED
580 * See that the long running process really only updates the principal
581 file unless it has missed a timespan during which something happened. If
582 nothing happened, it must notice even when it misses the timespan. DONE
584 * we must throw away the pathdb when we have reached the end of Z. From
585 that moment we can have a very small pathdb because the only reason for
586 a pathdb is that we know to ignore old records in old files. We won't
587 need this pathdb again before the next full pass over the data is
588 necessary and then we will rebuild it as we go along. DONE
590 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
592 * I think, Done::register_one is doing wrong in that it does not
593 conflate neighboring pieces. The covered() method cannot do this because
594 it has no recent_events array at hand. But register_one has it and could
595 do it and for some reason misses to do it (sometimes).
597 This means that the three tests I just wrote can probably not survive
598 because they test with an already broken Done structure.
600 The art now is to detect how it happens, then to reproduce, then write a
603 So from the logfile this is what happens: we have a good interval with
604 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
605 top of it. Locally we now mirror F2 and open a new done interval for it.
606 Then we mirror F1 but this time with the timestamp T1b. And when we then
607 try to close the gap, we do not find T1 but instead something older. We
608 should gladly accept this older piece and this would fix this bug.
612 * bug to fix: when the 1h file changes while rmirror is running, we do
613 correctly sync the new files but never switch to the 6h file but rather
614 stay in a rather quick loop that fetches the 1h file again and again.
616 Is it possible that we initialize a new object? Or does
617 get_remote_recentfile_as_tempfile overwrite something in myself?
619 Want a new option: _runstatusfile => $file which frequently dumps the
620 state of all recentfiles to a file.
624 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
626 * Todo: now teach update to verify the timestamp is about to write
627 against the previous and use _increase_a_bit if it doesn't comply with
628 strict monotony. DONE
630 * The problem of rounding. So far perl's default precision was
631 sufficient. One day it won't be. FakeFloat has an easy job when it is
632 only reading and other machines have written correctly. But when we want
633 to write a floating point number that is a bit larger than the other
634 one, then we need our own idea of precision.
636 Slaven said: just append a "1". This might be going towards the end of
637 usability too quickly. I'd like something that actually uses the decimal
638 system. Well, appending a 1 also does this but...
640 E.g. we have 1.0. nextup on this architecture is starting with
641 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
642 taken the 1.0000000000000003 and the next user comes and the time tells
643 him 1.0 again. He has to beat my number without stepping over the
644 nextup. This is much less space than I had when I chose 1,2,3.
646 What is also irritating is that nextup is architecture dependent. The
647 128 bit guy must choose very long numbers to fit in between whereas the
648 other one with 16 bit uses larger steps. But then the algorithm is the
649 same for both, so that would be a nice thing.
651 I see two situation where we need this. One is when Time::HiRes returns
652 us a value that is <= the last entry in our recentfile. In this case
653 (let's call it the end-case) we must fill the region between that number
654 and the next higher native floating point number. The other is when we
655 inject an old file into an old recentfile (we would then also set a new
656 dirtymark). We find the integer value already taken and need a slightly
657 different one (let's call it the middle-case). The difference between
658 the two situations is that the next user will want to find something
659 higher than my number in the end-case and something lower than my number
662 So I suggest we give the function both a value and an upper bound and it
663 calculates us a primitive middle. The upper bound in the middle-case is
664 the next integer. The upper bound on the end-case is the nextup floating
665 point number. But the latter poses another problem: if we have occupied
666 the middle m between x and nextup(x), then the nextup(m) will probably
667 not be the same as nextup(x) because some rounding will take place
668 before the nextup is calculated and when the rounding reaches the
669 nextup(x), we will end up at nextup(nextup(x)).
671 So we really need to consider the nextup and the nextdown from there and
672 then the middle and that's the number we may approach asymptotically.
675 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
677 * consider deprecating the use of RECENT.recent as a symlink. It turns
678 out to need extra hoops with the rsync options and just isn't worth it.
679 Or maybe these extra hoops are needed anyway for the rest of the tree?
680 Nope, can't be the case because not all filesystems support symlinks.
682 But before doing the large step, I'll deprecate the call of
683 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
684 only introduced to resolve RECENT.recent and complicates the routine far
685 beyond what it deserves.
687 DONE. Won't deprecate RECENT.recent, just moved its handling to the
690 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
692 * I think it's a bug that the rsync_option links must be set to true in
693 order to support RECENT.recent and that nobody cares to set it
694 automatically. Similar for ignore_link_stat_errors.
696 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
698 * Todo: find all todos together and make a plan what is missing for a
701 - verifytree or something like that. fsck maybe.
703 - rersyncrecent, the script itself? What it do?
705 - a way to only mirror the recentfiles without mirroring the whole
706 remote system such that people can decide to mirror only partially see
707 also 2008-08-30. .shadow-xxx directory? this also needed for a
708 filesystem that is still incomplete and might need the mirrorfiles for
711 - long living objects that mirror again and again. Inject something
712 into ta, see how it goes over to tb.
714 - how do we continue filling up the DONE system when we use an object
715 for the second time? "fully covered" and "uptodate" or new terminology.
717 - overview called on the wrong file should be understandable
719 - the meta data field that must change when we fake something up so that
720 the downstream people know they have to re-fetch everything.
722 - how tolerant are we against missing files upstream? how do we keep
723 track? there are legitimate cases where we did read upstream index right
724 before a file got deleted there and then find that file as new and want
725 it. There are other cases that are not self healing and must be tracked
728 - how, exactly, do we have to deal with deletes? With rsync errors?
730 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
731 authors) failed: No such file or directory (2)
733 The file above is a delete in 1h and a new in file 1M and the
734 delete in the locally running rmirror did not get propagated to the 1M
735 object. Bug. And the consequence is a standstill.
737 It seems that a slave that works with a file below the principal needs
738 to merge things all the way up to get rid of later deletes. Or keep
739 track of all deletes and skip them later. So we need a trackdeletes.pm
740 similar to the done.pm?
742 see also 2008-08-20 about spurious deletes that really have no add
743 counterpart and yet they are not wrong.
745 - consider the effect when resyncing the recentfile takes longer than
746 the time per loop. Then we never rsync any file. We need to diagnose
747 that and force an increase of that loop time. But when we later are fast
748 enough again because the net has recovered, then we need to switch back
749 to original parameters. ERm, no, it's enough to keep syncing at least
750 one file before refetching an index file.
752 - remember to verify that no temp files are left lying around and the
755 - status file for not long running jobs that want to track upstream with
758 - revisit all XXX _float areas and study Sub::Exporter DONE
760 - persistent DB even though we just said we do not need it. Just for
761 extended capabilities and time savings when, for example, upstream
762 announces a reset and we get new recentfiles and could then limit
763 ourselves to a subset of files (those that have a changed epoch) in a
764 first pass and would only then do the loop to verify the rest. Or
767 * Todo: aggregate files should know their feed and finding the principal
768 should be done stepwise. (?)
770 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
771 left locks around. DONE
773 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
775 * maybe extend the _overview so that it always says if and where the
776 last file is in the next file and where the next event in the next rf
777 would lie. No, don't like this anymore. REJECT
779 * take the two new redundant tests out again, only the third must
782 * Todo: add a sanity check if the merged structure is really pointing to
783 a different rf and that this different rf is larger. DONE
785 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
787 * now test, if they are overlapping. And test if there is a file in the
788 next rf that would fit into this rf's interval.
790 1h 1222324012.8474 1222322541.7963 0.4086
791 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
792 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
793 1W 1222313218.3626 1221708477.5829 167.9835
795 I suspect that somebody writes a merged timestamp without having merged
796 and then somebody else relies on it.
798 If aggregate is running, the intervals must not be extravagated, if it
799 is not running, there must not be bounds, the total number of events in
800 the system must be counted and must be controlled throughout the tests.
801 That the test required the additional update was probably nonsense,
802 because aggregate can cut pieces too. FIXED & DONE
804 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
806 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
809 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
811 * Most apparent bug at the moment is that the recentfiles are fetched
812 too often. Only the principal should be fetched and if it has not
813 changed, the others should not be refetched. ATM I must admit that I'm
814 happy that we refetch more often than needed because I can more easily
815 fix bugs while the thing is running.
817 * Let's say, 1220474966.19501 is a timestamp of a file that is already
818 done but the done system does not know about it. The reason for the
819 failure is not known and we never reach the status uptodate because of
820 this. We must get over it.
822 Later it turns out that the origin server had a bug somewhere.
823 1220474966.19042 came after 1220474966.19501. Or better: it was in the
824 array of the recentfile one position above. The bug was my own.
826 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
828 * There is the race condition where the server does a delete and the
829 slave does not yet know and then tries to download it because he sees
830 the new. So for this time window we must be more tolerant against
831 failure. If we cannot download a file, we should just skip it and should
832 not retry immediately. The whole system should discover the lost thing
833 later. Keeping track with the DONE system should really be a no brainer.
835 But there is something more: the whole filesystem is a database and the
836 recentfiles are one possible representation of it. It's a pretty useful
837 representation I think that's why I have implemented something around
838 it. But for strictly local operation it has little value. For local
839 operation we would much rather have a database. So we would enter every
840 recentfile reading and every rsync operation and for every file the last
841 state change and what it leads to. Then we would always ignore older
842 records without the efforts involved with recentfiles.
844 The database would have: path,recentepoch,rsyncedon,deletedon
846 Oh well, not yet clear where this leads to.
848 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
850 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
851 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
853 * statusfile, maybe only on demand, alone to have a sharp debugging
854 tool. It is locked and all recentfiles dump themselves into it and we
855 can build a viewer that lets us know where we stand and what's inside.
857 * remember: only the principal recentfile needs expiration, all others
858 shall be expired by principal if it discovers that something has move
861 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
863 * Always check if we stringify to a higher value than in the entry
866 * And in covered make an additional check if we would be able to see a
867 numerical difference between the two numbers and if we can't then switch
868 to a different, more expensive algorithm. Do not want to be caught by
869 floating surprises. DONE
871 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
873 * caching has several aspects here: we can cache the interval of the
874 recentfile which only will change when the mtime of the file changes. We
875 must re-mirror the recentfile when its ttl has expired. Does have_read
876 tell you anything? It counts nothing at all. Only the mtime is
877 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
878 thing the have_mirrored because it is unrelated to the mtime.
880 * Robustness of floating point calculations! I always thought that the
881 string calculated by the origin server for the floating representation
882 of the epoch time is just a string. When we convert it to a number and
883 later back to a string, the other computer might come to a different
884 conclusion. This must not happen, we want to preserve it under any
885 circumstances. I will have to write tests with overlong sequences that
886 get lost in arithmetic and must see if all still works well. DONE
888 But one fragile point remains: if one host considers a>b and the other
889 one considers them == but no eq. To prevent this, we must probably do
890 some extra homework. DONE
892 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
894 * the concept of tracking DONE needs an object per recentfile that has
895 something like these methods:
897 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
899 covered() register() covered()
901 The unclear thing is how we translate points in time into intervals. We
902 could pass a reference to the current recent_events array when running
903 we_have(xxx) and let the DONE object iterate over it such that it only
904 has to store a list of intervals that can melt into each other. Ah, even
905 passing the list together with a list of indexes seems feasiable.
907 Or maybe ask for the inverted list?
909 Whenever the complete array is covered by the interval we say we are
910 fully covered and if the recentfile is not expired, we are uptodate.
912 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
914 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
916 * need a way to "return" the next entry after the end of a list. When
917 the caller says "before" or "after" we would like to know if he could
918 cover that interval/threshold or not because this influences the effect
919 of a newer timestamp of that recentfile. DONE with $opt{info}.
921 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
923 * one of the next things to tackle: the equivalent of csync2 -TIXU.
925 loop implies tixu (?). Nope, something like --statefile decides. Per
928 T test, I init, X including removals, U nodirtymark
930 So we have no concept of dirtymarks, we only trust that since we are
931 running we have observed everything steadily. But people will not let
932 this program run forever so we must consider both startup penalty and
933 book keeping for later runs. We keep this for later. For now we write a
934 long running mirror that merges several intervals.
936 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
938 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
941 * rersyncrecent, the script: default to one week. The name of the switch
942 is --after. Other switches? --loop!
944 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
946 * need a switch --skip-deletes (?)
948 * need a switch --enduser that tells us that the whole tempfile
949 discipline is not needed when there is no downstream user. (?)
951 Without this switch we cannot have a reasonable recent.pl that just
952 displays the recent additions. Either we accept to download everything.
953 Or we download temporary files without the typical rsync protocol
956 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
957 File::Temp::tempdir, this might be a win.
959 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
961 * apropos missing: we have no push, we never know the downstream
962 servers. People who know their downstream hosts and want to ascertain
963 something will want additional methods we have never thought about, like
964 update or delete a certain file.
966 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
968 * tempted to refactor rmirror into resolve_symlink, localize, etc.
969 Curious if rsync_options=links equal 0 vs. 1 will make the expected
972 * rsync options: it's a bit of a pain that we usually need several rsync
973 options, like compress, links, times, checksum and that there is no
974 reasonable default except the original rsync default. I think wee can
975 safely assume that the rsync options are shared between all recentfile
976 instances within one recent tree.
978 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
980 * deletes: if a delete follows an add quickly enough it may happen that
981 a downstream mirror did not see the add at all! It seems this needs to
982 be mentioned somewhere. The point here is that even if the downstream is
983 never missing the principal timeframe it may encounter a "delete" that
984 has no complimentary "add" anywhere.
986 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
988 * I suspect the treat of metadata is incorrect during read or something.
989 The bug that I am watching is that between 06:08 and 06:09 the 6h file
990 contained more than 6 hours worth of data. At 06:08 we merged into the
991 1d file. We need to take snapshots of the 6h file over the course of an
992 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
995 Much worse: watching the 1h file: right at the moment (at 06:35) it
996 covers 1218867584-1219120397 which is 70 hours.
998 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
999 08:19:44 2008, that is when I checked out last time, so it seems to be
1000 aggregating and never truncating?
1002 No, correct is: it is never truncating; but wrong is: it is aggregating.
1003 It does receive a lot of events from time to time from a larger file.
1004 Somehow a large file gets merged into the small one and because the
1005 "meta/merged" attribute is missing, nobody is paying attention. I
1006 believe that I can fix this by making sure that metadata are honoured
1007 during read. DONE and test adjusted.
1009 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1011 * grand renaming plan
1013 remotebase => remoteroot to fit well with localroot DONE
1014 local_path() => localroot seems to me should already work DONE
1015 recentfile_basename => rfilename no need to stress it has no slash DONE
1017 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1018 kern? It would anyway need a deprecation cycle because it is an
1019 important constructor.
1021 * I like the portability that Data::Serializer brings us but the price
1022 is that some day we might find out that it is slowing us a bit. We'll
1025 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1027 * should we not enter the interval of the principal (or the interval of
1028 the merging file?) in every aggregated/merged file?
1030 * we should aim at a first release and give up on thinking about
1031 sanitizing stuff and zloop. Let's just admit that a full traditional
1032 rsync is the only available sanitizer ATM. Otherwise it's complicated
1033 stuff: sanitizing on the origin server, sanitizing on the slaves,
1034 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1035 get the basics out before this becomes a major cause for mess.
1037 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1039 * On OSes not supporting symlinks we expect that RECENT.recent contains
1040 the contents of the principal recentfile. Actually this is identical on
1041 systems supporting symlinks. Simple, what follows from that is that we
1042 need to keep the serializer in the metadata because we cannot read it
1043 from the filename, doesn't it? Of course not. It's a chicken and egg
1044 problem. This leaves us with the problem to actually parse the
1045 serialized data to find out in which format it is. So who can do the 4
1046 or 5 magics we wanted to support? File::LibMagic?
1048 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1050 * remotebase and recentfile_basename are ugly names. Now that we need a
1051 word for the shortest/principal/driving recentfile too we should do
1054 localroot is good. rfile is good. local_path() is bad, local_path($path)
1055 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1058 Up to now remotebase was the string that described the remote root
1059 directory in rsync notation, like pause.perl.org::authors. And
1060 recentfile_basename was "RECENT-1h.yaml".
1062 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1064 * The test that was added in today's checkin is a good start for a test
1065 of rmirror. We should have more methods in Recent.pm: verify,
1066 addmissingfiles. We should verify the current tree, then rmirror it and
1067 then verifytree the copy. We could then add some arbitrary file and let
1068 it be discovered by addmissingfiles, then rmirror again and then
1069 verifytree the copy again.
1071 Then we could start stealing from csync2 sqlite database [no port to
1072 OSX!] and fill a local DB. And methods to compare the database with the
1073 recentfiles. Our strength is that in principle we could maintain state
1074 with a single float. We have synced up to 1234567890.123456. If the Z
1075 file does not add new files all we have to do is mirror the new ones and
1078 This makes it clear that we should extend current protocol and declare
1079 that we cheat when we add files too late, just to help the other end
1080 keeping track. Ah yes, that's what was meant when zloop was mentioned
1083 Maybe need to revisit File::Mirror to help me with this task.
1085 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1087 * There must be an allow-me-to-truncate flag in every recentfile.
1088 Without it one could construct a sequence of updates winning the locking
1089 battle against the aggregator. Only if an aggregator has managed to
1090 merge data over to the next level, truncating can be allowed. DONE with
1093 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1095 * We should probably guarantee that no duplicates enter the aggregator
1098 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1100 * To get merge operation faster would need a good benchmark test. What
1101 02 spits out isn't reliable enough and is dominated by many other
1104 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1105 Author: Andreas J. Koenig <k@k75.(none)>
1106 Date: Sat Aug 2 07:58:04 2008 +0200
1110 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1111 Author: Andreas J. Koenig <k@k75.(none)>
1112 Date: Sat Aug 2 11:40:29 2008 +0200
1114 there was a lot of trying but the effect is hardly measurable with
1117 * overhead of connecting seems high. When setting
1118 max_files_per_connection to 1 we see that.
1120 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1122 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1124 25138.2215600014/3600 = 6.98283932222261
1126 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1127 be a bug in the aggregator. FIXED
1129 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1131 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1132 we have it, can we mirror it, mirror it!
1134 I fear this needs a new class which might be called
1135 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1136 and treat them as an entity. I realize that a single recentfile may be
1137 sufficient for certain tasks and that it is handy for the low level
1138 programmer but it is not nice to use. If there is a delete in the 1h
1139 file then the 6h file still contains it. Seekers of the best information
1140 need to combine at least some of the recentfiles most of the time.
1142 There is the place for the Z loop!
1144 But the combination is something to collect in a database, isn't it. Did
1145 csync2 just harrumph?
1147 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1149 * it just occurred to me that hosts in the same mirroring pool could
1150 help out each other even without rewriting the recentfile. Just fetch
1151 the stuff to mirror from several places, bingo. But that's something
1152 that should rather live in a separate package or in rsync directly.
1154 * cronjobs are unsuited because with ntp they would all come at the full
1155 minute and disturb each other. Besides that I'd hate to have a backbone
1156 with more than a few seconds latency.
1158 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1160 * a second rsync server with access control for PAUSE. Port? 873 is the
1161 standard port, let's take 8873.
1163 * if there were a filesystem based on this, it would have a slow access
1164 to inexistent files. It would probably provide wrong readdir (only based
1165 on current content) or also a slow one (based on a recentfile written
1166 after the call). But it would provide fast access to existing files. Or
1167 one would deliberately allow slightly blurred answers based on some
1168 sqlite reflection of the recentfiles.
1170 * todo: write a variant of mirror() that combines two or more
1171 recentfiles and treats them like one
1173 * todo: signal handler to remove the tempfile
1175 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1177 * now that we have the symlink I forgot how it should be used in
1180 * the z loop: add missing files to Z file. Just append them (instead of
1181 prepending). So one guy prepends something from the Y file from time to
1182 time and another guy appends something rather frequently. Collecting
1183 pond. When Y merges into Z, things get epoch and the collecting pond
1184 gets smaller. What exactly are "missing files"?
1186 take note of current epoch of the alpha file, let's call it the
1189 find all files on disk
1191 remove all files registered in the recentworld up to recent-ts
1193 remove all files that have been deleted after recent-ts according to
1196 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1198 * rersyncrecent might be a cronjob with a (locked) state file which
1199 contains things like after and maybe last z sync or such?
1201 rrr-mirror might be an alternative name but how would we justify the
1202 three Rs when there is no Re-Rsync-Recent?
1204 With the --loop parameter it is an endless loop, without it is no loop.
1205 At least this is simple.
1207 * todo: new accssor z-interval specifies how often the Z file is updated
1208 against the filesystem. We probably want no epoch stamp on these
1209 entries. And we want to be able to filter the entries (e.g. no
1210 by-modules and by-category tree)
1212 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1214 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1215 into Z. We do not need the other files filled up with filesystem
1218 * need interface to query for a file in order to NOT call update on
1219 PAUSE a second time within a short time.
1221 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1223 * recommended update interval? Makes no sense, is different for
1230 change-log-default-name: "Todo"