1 2009-03-24 Andreas J. Koenig <andk@cpan.org>
5 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
9 2009-03-22 Andreas J. Koenig <andk@cpan.org>
11 * k81 is again client of k75 with different parameters.
13 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
14 about it all the time and nobody reacts accordingly; (2) lots of temp
15 files get created and not removed; culprit the new call to
16 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
17 caller has to remove the tempfile after use.
19 Need the drawing board.
21 PARTLY FIXED: retracting the idea to call
22 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
23 and unseed such that uptodate now means "we have mirrored this
24 recentfile's bunch of files" and seeded means "somebody believes the
25 index file for this rf needs be refreshed". This seems to work now.
27 Still collecting temp files that nobody cares to depollute.
29 2009-03-21 Andreas J. Koenig <andk@cpan.org>
31 * Bug?: should it be harder than it is atm to set the timestamp to the
34 * bug with native integers:
38 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
42 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
45 Native integer broke when native math was turned off. FIXED
47 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
48 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
51 Yes, records out of order:
54 447004 epoch: 995885533
55 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
58 447008 epoch: 995890358
59 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
62 447012 epoch: 995892221
63 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
66 FIXED with sanity check and later with the integer fix.
68 * Bug: want the index files in a .recent directory
70 * Bug: lots of dot files are not deleted in time
72 * possible test case: can a delete change the timestamp? This would
73 probably break the order of events.
75 2009-03-20 Andreas J. Koenig <andk@cpan.org>
77 * 1233701831.34486 what's so special about this number/string? It
79 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
80 atm lives in Y,Q, and Z.
82 It is the first entry after id/--skip-locking which has timestamp
83 1234164228.11325 which represents a file that doesn't exist anymore.
87 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
88 _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
89 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
90 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
91 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
92 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
93 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
94 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
95 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
96 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
97 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
98 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
99 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
101 FIXED, it was the "--skip-locking" file where manual intervention was participating
103 * bug on the mirroring slave: when the dirtymark gets increased we
104 probably do not reset the done intervals. The mirrorer stays within
105 tight bounds where it tries to sync with upstream and never seems to
106 finish. In the debugging state file I see lots of identical intervals
107 that do not get collapsed. When I restart the mirrorer it dies with:
109 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
110 _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
111 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
112 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
113 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
114 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
115 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
116 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
117 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
118 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
120 and debugging stands at
126 epoch: 1237400802.5789
127 path: id/J/JO/JOHND/CHECKSUMS
130 epoch: 1237400807.97514
131 path: id/M/MI/MIYAGAWA/CHECKSUMS
134 epoch: 1237400817.94363
135 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
143 and it is reproducable.
145 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
146 pause has a fresh one.
148 FIXED, it was the third anded term in each of the ifs in the IV block in
149 _register_one: with that we make sure that we do not stamp on valuable
152 2009-03-17 Andreas J. Koenig <andk@cpan.org>
154 * done: verified the existence of the floating point bug in bleadperl
155 and verified that switching from YAML::Syck to YAML::XS does not resolve
158 BTW, the switch was doable with
160 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
162 and should be considered as a separate TODO
164 * todo: integrate a dirty update with two aggregate calls before
165 unlocking for frictionless dirtying
167 * todo: start the second rsync daemon on pause
169 * todo: move index files to .recent: this cannot simply be done by
170 setting filenameroot to .recent/RECENT. Other parts of the modules rely
171 on the fact that dirname(recentfile) is the root of the mirrored tree.
173 2009-03-16 Andreas J. Koenig <andk@cpan.org>
175 * What was the resolution of the mirror.pl delete hook bug? Do we call
176 the delete hook when pause removes a file from MUIR?
178 * Today on pause: Updating 2a13fba..29f284d and installing it for
179 /usr/local/perl-5.10.0{,-RC2}
181 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
182 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
184 2009-03-15 Andreas J. Koenig <andk@cpan.org>
186 * currently recent_events has the side effect of setting dirtymark
187 because it forces a read on the file. That should be transparent, so
188 that the dirtymark call always forces a cache-able(?) read.
190 * The bug below is -- after a lot of trying -- not reproducible on a
191 small script, only in the large test script. The closest to the output
197 my $x = "01237123229.8814";
202 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
204 $l = "1237123231.22458";
205 $r = "1237123231.22458";
207 Devel::Peek::Dump $l;
208 Devel::Peek::Dump $r;
209 Devel::Peek::Dump $x = $l <=> $r;
213 The checked in state at c404a85 fails the test with my
214 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
215 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
216 in the test script always consistent.
218 * Todo: write a test that inserts a second dirty file with an already
219 existing timestamp. DONE
221 * Bug in perl 5.10 on my 64bit box:
223 DB<98> Devel::Peek::Dump $l
224 SV = PVMG(0x19e0450) at 0x142a550
226 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
228 NV = 1237123231.22458
229 PV = 0x194ce70 "1237123231.22458"\0
233 DB<99> Devel::Peek::Dump $r
234 SV = PVMG(0x19e0240) at 0x142a3e8
236 FLAGS = (PADMY,POK,pPOK)
239 PV = 0x19ff900 "1237123231.22458"\0
243 DB<100> Devel::Peek::Dump $l <=> $r
244 SV = IV(0x19ea6e8) at 0x19ea6f0
246 FLAGS = (PADTMP,IOK,pIOK)
249 DB<101> Devel::Peek::Dump $l
250 SV = PVMG(0x19e0450) at 0x142a550
252 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
254 NV = 1237123231.22458
255 PV = 0x194ce70 "1237123231.22458"\0
259 DB<102> Devel::Peek::Dump $r
260 SV = PVMG(0x19e0240) at 0x142a3e8
262 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
264 NV = 1237123231.22458
265 PV = 0x19ff900 "1237123231.22458"\0
269 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
271 * Todo: reset "done" or "covered" and "minmax" after a dirty operation?
273 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
275 * $obj->merge ($other) needs to learn about equal epoch which may happen
276 since dirty_epoch intruded.
278 * Wontfix anytime soon: I think we currently do not support mkdir. Only
281 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
283 * Todo: continue working on update(...,$dirty_epoch). It must be
284 followed by a fast_aggregate!
286 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
288 * maybe we need a closest_entry or fitting_interval or something like
289 that. We want to merge an event into the middle of some recentfile.
290 First we do not know which file, then we do not know where to lock,
291 where to enter the new item, when and where to correct the dirtymark.
293 So my thought is we should first find which file.
295 Another part of my brain answers: what would happen if we would enter
296 the new file into the smallest file just like an ordinary new event,
297 just as an old event?
299 (1) we would write a duplicate timestamp? No, this would be easy to
302 (2) we would make the file large quickly? Yes, but so what? We are
303 changing the dirtymark, so are willing to disturb the downstream hosts.
305 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
307 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
311 https://rt.cpan.org/Ticket/Display.html?id=41199
313 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
315 Strangest output being HAKANARDO who managed to upload
317 Here is a better oneliner that includes also the first line of each
320 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
322 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
324 * memory leak in the syncher? It currently weighs 100M.
328 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
331 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
333 * lookup by epoch and by path and use this ability on the pause to never
334 again register a file twice that doesn't need it. Let's call it
337 * after the dirtymark is done: fill up recentfiles with fake (historic)
338 entries; fill up with individual corrections; algorithm maybe to be done
339 with bigfloat so that we can always place something in the middle
340 between two entries. Before we must switch to bigfloat we could try to
341 use Data::Float::nextup to get the.
343 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
346 * dirtymark now lives in Recentfile, needs to be used in rmirror.
348 * find out why the downloader died after a couple of hours without a net
349 connection. Write a test that survives the not-existence of the other
352 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
354 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
355 it makes less sense with tight coupling for secondary files. But asking
356 the server what the current epoch is might be cheaper on HTTP than on
357 rsync. (Needs to be evaluated)
359 * remove the 0.00 from the verbose overview in the Merged column in the
362 * write tests that expose the problems of the last few days: cascading
363 client/server roles, tight coupling for secondary RFs, deletes after
366 * Some day we might want to have policy options for the slave:
367 tight/loose/no coupling with upstream for secondary RFs. tight is what
368 we have now. loose would wait until a gap occurs that can be closed.
370 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
372 * revisit all $rfs->[$i+1] places if they now make sense still
374 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
376 * another bug is the fact that the mirror command deletes files before
377 it unhides the index file, thus confusing downstream slaves. We must not
378 delete before unhiding and must delete after unhiding. FIXED.
380 * new complication about the slave that is playing a server role.
381 Currently we mirror from newest to oldest with a hidden temporary file
382 as index. And when one file is finished, we unhide the index file.
383 Imagine the cascading server/slave is dead for a day. It then starts
384 mirroring again with the freshest thing and unhides the freshest index
385 file when it has worked through it. In that moment it exposes a time
386 hole. Because it now works on the second recentfile which is still
389 We currently do nothing special to converge after such a drop out. At
390 least not intentionally and robustly and thought through.
392 The algorithm we use to seed the next file needs quite a lot of more
393 robustness than it currently has. Something to do with looking at the
394 merged element of the next rf and when it has dropped off, we seed
395 immediately. And if it ramains dropped off, we seed again, of course.
397 Nope, looking from smaller to larger RFS we look at the merged element
398 of this RF and at the minmax/max element of the next RF. If that
399 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
402 And we need a public accessor seed and unseed or seeded. But not the mix
403 of public and private stuff that then is used behind the back.
405 And then the secondary* stuff must go.
407 And we must understand what the impact is on the DONE system. Can it go
408 unnoticed that there was a hole? And could the DONE system have decided
409 the hole is covered? This should be testable with three directories where
410 the middle stops working for a while. Done->merge is suspicious, we must
411 stop it from merging non-conflatable neighbors due to broken continuity.
415 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
417 * Slaven suggests to have the current epoch or the whole current
418 recentfile available from the HTTP server and take it away with
419 keepalive. This direction goes the granularity down to subseconds.
421 We might want to rewrite everything to factor out transport and allow
422 the whole thing to run via HTTP.
424 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
426 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
429 * smoker on k81 fetching from k75 to verify cascading works. See
430 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
432 * maybe the loop should wait for CHECKSUMS file after every upload. And
433 CPAN.pm needs to deal with timestamps in the future.
435 * do not forget the dirtymark!
437 Text: have a new flag on recentfiles with the meaning: if this
438 changes, you're required to run a full rsync over all the files. The
439 reason why we set it would probably be: some foul happened. we injected
440 files in arbitrary places or didn't inject them although they changed.
441 The content of the flag? Timestamp? The relation between the
442 recentfiles would have to be inheritance from the principal, because any
443 out of band changes would soon later propagate to the next recentfile.
445 By upping the flag often one can easily ruin the slaves.
447 last out of band change? dirtymark?
449 Anyway, this implies that we read a potentially existing recentfile
452 And it implies that we have an eventloop that keeps us busy in 2-3
453 cycles, one for current stuff (tight loop) and one for the recentfiles
454 (cascade when principal has changed), one for the old stuff after a
457 And it implies that the out-of-band change in any of the recentfiles
458 must have a lock on the principal file and there is the place to set the
461 * start a FAQ, especially quick start guide questions. Also to aid those
462 problematic areas where we have no good solution, like the "links"
465 * wish feedback when we are slow.
469 * Remove a few DEBUG statements.
471 * The multiple-rrr way of doing things needs a new option to rmirror,
472 like piecemeal or so. Not urgent because after the first pass through,
473 things run smoothely. It's only ugly during the first pass.
475 * I have the suspicion that the code is broken that decides if the
476 neighboring RF needs to be seeded. I fear when too much time has gone
477 between two calls (in our case more than one hour), it would not seed
478 the neighbor. Of course this will never be noticed, so we need a good
481 * local/localroot confusion: I currently pass both options but one must
484 * accounts for early birds on PAUSE rsync daemon.
486 * hardcoded 20 seconds
488 * who mirrors the index? DOING now.
490 * which CPAN mirrors offer rsync?
492 * visit all XXX, visit all _float places
494 * rename the pathdb stuff, it's too confusing. No idea how.
496 * rrr-inotify, backpan, rrr-register
498 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
500 * current bugs: the pathdb seems to get no reset, the seeding of the
501 secondaryttl stuff seems not to have an effect. Have helped myself with
502 a rand(10), need to fix this back. So not checked in. Does the rand
505 The rand thing helps. The secondaryttl stuff was in the wrong line,
508 The pathdb stuff was because I called either _pathdb or __pathdb on the
509 wrong object. FIXED now.
511 * It's not so beautiful if we never fetch the recentfiles that are not
512 the principal, even if this is correct behaviour. We really do not need
513 them after we have fetched the whole content.
515 OK, we want a switch for that: secondaryttl DONE
517 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
519 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
520 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
521 it lets 40 follow. FIXED
523 * See that the long running process really only updates the principal
524 file unless it has missed a timespan during which something happened. If
525 nothing happened, it must notice even when it misses the timespan. DONE
527 * we must throw away the pathdb when we have reached the end of Z. From
528 that moment we can have a very small pathdb because the only reason for
529 a pathdb is that we know to ignore old records in old files. We won't
530 need this pathdb again before the next full pass over the data is
531 necessary and then we will rebuild it as we go along. DONE
533 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
535 * I think, Done::register_one is doing wrong in that it does not
536 conflate neighboring pieces. The covered() method cannot do this because
537 it has no recent_events array at hand. But register_one has it and could
538 do it and for some reason misses to do it (sometimes).
540 This means that the three tests I just wrote can probably not survive
541 because they test with an already broken Done structure.
543 The art now is to detect how it happens, then to reproduce, then write a
546 So from the logfile this is what happens: we have a good interval with
547 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
548 top of it. Locally we now mirror F2 and open a new done interval for it.
549 Then we mirror F1 but this time with the timestamp T1b. And when we then
550 try to close the gap, we do not find T1 but instead something older. We
551 should gladly accept this older piece and this would fix this bug.
555 * bug to fix: when the 1h file changes while rmirror is running, we do
556 correctly sync the new files but never switch to the 6h file but rather
557 stay in a rather quick loop that fetches the 1h file again and again.
559 Is it possible that we initialize a new object? Or does
560 get_remote_recentfile_as_tempfile overwrite something in myself?
562 Want a new option: _runstatusfile => $file which frequently dumps the
563 state of all recentfiles to a file.
567 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
569 * Todo: now teach update to verify the timestamp is about to write
570 against the previous and use _increase_a_bit if it doesn't comply with
571 strict monotony. DONE
573 * The problem of rounding. So far perl's default precision was
574 sufficient. One day it won't be. FakeFloat has an easy job when it is
575 only reading and other machines have written correctly. But when we want
576 to write a floating point number that is a bit larger than the other
577 one, then we need our own idea of precision.
579 Slaven said: just append a "1". This might be going towards the end of
580 usability too quickly. I'd like something that actually uses the decimal
581 system. Well, appending a 1 also does this but...
583 E.g. we have 1.0. nextup on this architecture is starting with
584 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
585 taken the 1.0000000000000003 and the next user comes and the time tells
586 him 1.0 again. He has to beat my number without stepping over the
587 nextup. This is much less space than I had when I chose 1,2,3.
589 What is also irritating is that nextup is architecture dependent. The
590 128 bit guy must choose very long numbers to fit in between whereas the
591 other one with 16 bit uses larger steps. But then the algorithm is the
592 same for both, so that would be a nice thing.
594 I see two situation where we need this. One is when Time::HiRes returns
595 us a value that is <= the last entry in our recentfile. In this case
596 (let's call it the end-case) we must fill the region between that number
597 and the next higher native floating point number. The other is when we
598 inject an old file into an old recentfile (we would then also set a new
599 dirtymark). We find the integer value already taken and need a slightly
600 different one (let's call it the middle-case). The difference between
601 the two situations is that the next user will want to find something
602 higher than my number in the end-case and something lower than my number
605 So I suggest we give the function both a value and an upper bound and it
606 calculates us a primitive middle. The upper bound in the middle-case is
607 the next integer. The upper bound on the end-case is the nextup floating
608 point number. But the latter poses another problem: if we have occupied
609 the middle m between x and nextup(x), then the nextup(m) will probably
610 not be the same as nextup(x) because some rounding will take place
611 before the nextup is calculated and when the rounding reaches the
612 nextup(x), we will end up at nextup(nextup(x)).
614 So we really need to consider the nextup and the nextdown from there and
615 then the middle and that's the number we may approach asymptotically.
618 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
620 * consider deprecating the use of RECENT.recent as a symlink. It turns
621 out to need extra hoops with the rsync options and just isn't worth it.
622 Or maybe these extra hoops are needed anyway for the rest of the tree?
623 Nope, can't be the case because not all filesystems support symlinks.
625 But before doing the large step, I'll deprecate the call of
626 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
627 only introduced to resolve RECENT.recent and complicates the routine far
628 beyond what it deserves.
630 DONE. Won't deprecate RECENT.recent, just moved its handling to the
633 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
635 * I think it's a bug that the rsync_option links must be set to true in
636 order to support RECENT.recent and that nobody cares to set it
637 automatically. Similar for ignore_link_stat_errors.
639 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
641 * Todo: find all todos together and make a plan what is missing for a
644 - verifytree or something like that. fsck maybe.
646 - rersyncrecent, the script itself? What it do?
648 - a way to only mirror the recentfiles without mirroring the whole
649 remote system such that people can decide to mirror only partially see
650 also 2008-08-30. .shadow-xxx directory? this also needed for a
651 filesystem that is still incomplete and might need the mirrorfiles for
654 - long living objects that mirror again and again. Inject something
655 into ta, see how it goes over to tb.
657 - how do we continue filling up the DONE system when we use an object
658 for the second time? "fully covered" and "uptodate" or new terminology.
660 - overview called on the wrong file should be understandable
662 - the meta data field that must change when we fake something up so that
663 the downstream people know they have to re-fetch everything.
665 - how tolerant are we against missing files upstream? how do we keep
666 track? there are legitimate cases where we did read upstream index right
667 before a file got deleted there and then find that file as new and want
668 it. There are other cases that are not self healing and must be tracked
671 - how, exactly, do we have to deal with deletes? With rsync errors?
673 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
674 authors) failed: No such file or directory (2)
676 The file above is a delete in 1h and a new in file 1M and the
677 delete in the locally running rmirror did not get propagated to the 1M
678 object. Bug. And the consequence is a standstill.
680 It seems that a slave that works with a file below the principal needs
681 to merge things all the way up to get rid of later deletes. Or keep
682 track of all deletes and skip them later. So we need a trackdeletes.pm
683 similar to the done.pm?
685 see also 2008-08-20 about spurious deletes that really have no add
686 counterpart and yet they are not wrong.
688 - consider the effect when resyncing the recentfile takes longer than
689 the time per loop. Then we never rsync any file. We need to diagnose
690 that and force an increase of that loop time. But when we later are fast
691 enough again because the net has recovered, then we need to switch back
692 to original parameters. ERm, no, it's enough to keep syncing at least
693 one file before refetching an index file.
695 - remember to verify that no temp files are left lying around and the
698 - status file for not long running jobs that want to track upstream with
701 - revisit all XXX _float areas and study Sub::Exporter DONE
703 - persistent DB even though we just said we do not need it. Just for
704 extended capabilities and time savings when, for example, upstream
705 announces a reset and we get new recentfiles and could then limit
706 ourselves to a subset of files (those that have a changed epoch) in a
707 first pass and would only then do the loop to verify the rest. Or
710 * Todo: aggregate files should know their feed and finding the principal
711 should be done stepwise. (?)
713 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
714 left locks around. DONE
716 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
718 * maybe extend the _overview so that it always says if and where the
719 last file is in the next file and where the next event in the next rf
720 would lie. No, don't like this anymore. REJECT
722 * take the two new redundant tests out again, only the third must
725 * Todo: add a sanity check if the merged structure is really pointing to
726 a different rf and that this different rf is larger. DONE
728 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
730 * now test, if they are overlapping. And test if there is a file in the
731 next rf that would fit into this rf's interval.
733 1h 1222324012.8474 1222322541.7963 0.4086
734 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
735 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
736 1W 1222313218.3626 1221708477.5829 167.9835
738 I suspect that somebody writes a merged timestamp without having merged
739 and then somebody else relies on it.
741 If aggregate is running, the intervals must not be extravagated, if it
742 is not running, there must not be bounds, the total number of events in
743 the system must be counted and must be controlled throughout the tests.
744 That the test required the additional update was probably nonsense,
745 because aggregate can cut pieces too. FIXED & DONE
747 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
749 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
752 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
754 * Most apparent bug at the moment is that the recentfiles are fetched
755 too often. Only the principal should be fetched and if it has not
756 changed, the others should not be refetched. ATM I must admit that I'm
757 happy that we refetch more often than needed because I can more easily
758 fix bugs while the thing is running.
760 * Let's say, 1220474966.19501 is a timestamp of a file that is already
761 done but the done system does not know about it. The reason for the
762 failure is not known and we never reach the status uptodate because of
763 this. We must get over it.
765 Later it turns out that the origin server had a bug somewhere.
766 1220474966.19042 came after 1220474966.19501. Or better: it was in the
767 array of the recentfile one position above. The bug was my own.
769 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
771 * There is the race condition where the server does a delete and the
772 slave does not yet know and then tries to download it because he sees
773 the new. So for this time window we must be more tolerant against
774 failure. If we cannot download a file, we should just skip it and should
775 not retry immediately. The whole system should discover the lost thing
776 later. Keeping track with the DONE system should really be a no brainer.
778 But there is something more: the whole filesystem is a database and the
779 recentfiles are one possible representation of it. It's a pretty useful
780 representation I think that's why I have implemented something around
781 it. But for strictly local operation it has little value. For local
782 operation we would much rather have a database. So we would enter every
783 recentfile reading and every rsync operation and for every file the last
784 state change and what it leads to. Then we would always ignore older
785 records without the efforts involved with recentfiles.
787 The database would have: path,recentepoch,rsyncedon,deletedon
789 Oh well, not yet clear where this leads to.
791 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
793 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
794 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
796 * statusfile, maybe only on demand, alone to have a sharp debugging
797 tool. It is locked and all recentfiles dump themselves into it and we
798 can build a viewer that lets us know where we stand and what's inside.
800 * remember: only the principal recentfile needs expiration, all others
801 shall be expired by principal if it discovers that something has move
804 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
806 * Always check if we stringify to a higher value than in the entry
809 * And in covered make an additional check if we would be able to see a
810 numerical difference between the two numbers and if we can't then switch
811 to a different, more expensive algorithm. Do not want to be caught by
812 floating surprises. DONE
814 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
816 * caching has several aspects here: we can cache the interval of the
817 recentfile which only will change when the mtime of the file changes. We
818 must re-mirror the recentfile when its ttl has expired. Does have_read
819 tell you anything? It counts nothing at all. Only the mtime is
820 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
821 thing the have_mirrored because it is unrelated to the mtime.
823 * Robustness of floating point calculations! I always thought that the
824 string calculated by the origin server for the floating representation
825 of the epoch time is just a string. When we convert it to a number and
826 later back to a string, the other computer might come to a different
827 conclusion. This must not happen, we want to preserve it under any
828 circumstances. I will have to write tests with overlong sequences that
829 get lost in arithmetic and must see if all still works well. DONE
831 But one fragile point remains: if one host considers a>b and the other
832 one considers them == but no eq. To prevent this, we must probably do
833 some extra homework. DONE
835 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
837 * the concept of tracking DONE needs an object per recentfile that has
838 something like these methods:
840 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
842 covered() register() covered()
844 The unclear thing is how we translate points in time into intervals. We
845 could pass a reference to the current recent_events array when running
846 we_have(xxx) and let the DONE object iterate over it such that it only
847 has to store a list of intervals that can melt into each other. Ah, even
848 passing the list together with a list of indexes seems feasiable.
850 Or maybe ask for the inverted list?
852 Whenever the complete array is covered by the interval we say we are
853 fully covered and if the recentfile is not expired, we are uptodate.
855 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
857 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
859 * need a way to "return" the next entry after the end of a list. When
860 the caller says "before" or "after" we would like to know if he could
861 cover that interval/threshold or not because this influences the effect
862 of a newer timestamp of that recentfile. DONE with $opt{info}.
864 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
866 * one of the next things to tackle: the equivalent of csync2 -TIXU.
868 loop implies tixu (?). Nope, something like --statefile decides. Per
871 T test, I init, X including removals, U nodirtymark
873 So we have no concept of dirtymarks, we only trust that since we are
874 running we have observed everything steadily. But people will not let
875 this program run forever so we must consider both startup penalty and
876 book keeping for later runs. We keep this for later. For now we write a
877 long running mirror that merges several intervals.
879 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
881 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
884 * rersyncrecent, the script: default to one week. The name of the switch
885 is --after. Other switches? --loop!
887 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
889 * need a switch --skip-deletes (?)
891 * need a switch --enduser that tells us that the whole tempfile
892 discipline is not needed when there is no downstream user. (?)
894 Without this switch we cannot have a reasonable recent.pl that just
895 displays the recent additions. Either we accept to download everything.
896 Or we download temporary files without the typical rsync protocol
899 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
900 File::Temp::tempdir, this might be a win.
902 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
904 * apropos missing: we have no push, we never know the downstream
905 servers. People who know their downstream hosts and want to ascertain
906 something will want additional methods we have never thought about, like
907 update or delete a certain file.
909 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
911 * tempted to refactor rmirror into resolve_symlink, localize, etc.
912 Curious if rsync_options=links equal 0 vs. 1 will make the expected
915 * rsync options: it's a bit of a pain that we usually need several rsync
916 options, like compress, links, times, checksum and that there is no
917 reasonable default except the original rsync default. I think wee can
918 safely assume that the rsync options are shared between all recentfile
919 instances within one recent tree.
921 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
923 * deletes: if a delete follows an add quickly enough it may happen that
924 a downstream mirror did not see the add at all! It seems this needs to
925 be mentioned somewhere. The point here is that even if the downstream is
926 never missing the principal timeframe it may encounter a "delete" that
927 has no complimentary "add" anywhere.
929 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
931 * I suspect the treat of metadata is incorrect during read or something.
932 The bug that I am watching is that between 06:08 and 06:09 the 6h file
933 contained more than 6 hours worth of data. At 06:08 we merged into the
934 1d file. We need to take snapshots of the 6h file over the course of an
935 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
938 Much worse: watching the 1h file: right at the moment (at 06:35) it
939 covers 1218867584-1219120397 which is 70 hours.
941 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
942 08:19:44 2008, that is when I checked out last time, so it seems to be
943 aggregating and never truncating?
945 No, correct is: it is never truncating; but wrong is: it is aggregating.
946 It does receive a lot of events from time to time from a larger file.
947 Somehow a large file gets merged into the small one and because the
948 "meta/merged" attribute is missing, nobody is paying attention. I
949 believe that I can fix this by making sure that metadata are honoured
950 during read. DONE and test adjusted.
952 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
954 * grand renaming plan
956 remotebase => remoteroot to fit well with localroot DONE
957 local_path() => localroot seems to me should already work DONE
958 recentfile_basename => rfilename no need to stress it has no slash DONE
960 filenameroot??? Doesn't seem too bad to me today. Maybe something like
961 kern? It would anyway need a deprecation cycle because it is an
962 important constructor.
964 * I like the portability that Data::Serializer brings us but the price
965 is that some day we might find out that it is slowing us a bit. We'll
968 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
970 * should we not enter the interval of the principal (or the interval of
971 the merging file?) in every aggregated/merged file?
973 * we should aim at a first release and give up on thinking about
974 sanitizing stuff and zloop. Let's just admit that a full traditional
975 rsync is the only available sanitizer ATM. Otherwise it's complicated
976 stuff: sanitizing on the origin server, sanitizing on the slaves,
977 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
978 get the basics out before this becomes a major cause for mess.
980 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
982 * On OSes not supporting symlinks we expect that RECENT.recent contains
983 the contents of the principal recentfile. Actually this is identical on
984 systems supporting symlinks. Simple, what follows from that is that we
985 need to keep the serializer in the metadata because we cannot read it
986 from the filename, doesn't it? Of course not. It's a chicken and egg
987 problem. This leaves us with the problem to actually parse the
988 serialized data to find out in which format it is. So who can do the 4
989 or 5 magics we wanted to support? File::LibMagic?
991 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
993 * remotebase and recentfile_basename are ugly names. Now that we need a
994 word for the shortest/principal/driving recentfile too we should do
997 localroot is good. rfile is good. local_path() is bad, local_path($path)
998 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1001 Up to now remotebase was the string that described the remote root
1002 directory in rsync notation, like pause.perl.org::authors. And
1003 recentfile_basename was "RECENT-1h.yaml".
1005 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1007 * The test that was added in today's checkin is a good start for a test
1008 of rmirror. We should have more methods in Recent.pm: verify,
1009 addmissingfiles. We should verify the current tree, then rmirror it and
1010 then verifytree the copy. We could then add some arbitrary file and let
1011 it be discovered by addmissingfiles, then rmirror again and then
1012 verifytree the copy again.
1014 Then we could start stealing from csync2 sqlite database [no port to
1015 OSX!] and fill a local DB. And methods to compare the database with the
1016 recentfiles. Our strength is that in principle we could maintain state
1017 with a single float. We have synced up to 1234567890.123456. If the Z
1018 file does not add new files all we have to do is mirror the new ones and
1021 This makes it clear that we should extend current protocol and declare
1022 that we cheat when we add files too late, just to help the other end
1023 keeping track. Ah yes, that's what was meant when zloop was mentioned
1026 Maybe need to revisit File::Mirror to help me with this task.
1028 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1030 * There must be an allow-me-to-truncate flag in every recentfile.
1031 Without it one could construct a sequence of updates winning the locking
1032 battle against the aggregator. Only if an aggregator has managed to
1033 merge data over to the next level, truncating can be allowed. DONE with
1036 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1038 * We should probably guarantee that no duplicates enter the aggregator
1041 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1043 * To get merge operation faster would need a good benchmark test. What
1044 02 spits out isn't reliable enough and is dominated by many other
1047 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1048 Author: Andreas J. Koenig <k@k75.(none)>
1049 Date: Sat Aug 2 07:58:04 2008 +0200
1053 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1054 Author: Andreas J. Koenig <k@k75.(none)>
1055 Date: Sat Aug 2 11:40:29 2008 +0200
1057 there was a lot of trying but the effect is hardly measurable with
1060 * overhead of connecting seems high. When setting
1061 max_files_per_connection to 1 we see that.
1063 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1065 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1067 25138.2215600014/3600 = 6.98283932222261
1069 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1070 be a bug in the aggregator. FIXED
1072 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1074 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1075 we have it, can we mirror it, mirror it!
1077 I fear this needs a new class which might be called
1078 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1079 and treat them as an entity. I realize that a single recentfile may be
1080 sufficient for certain tasks and that it is handy for the low level
1081 programmer but it is not nice to use. If there is a delete in the 1h
1082 file then the 6h file still contains it. Seekers of the best information
1083 need to combine at least some of the recentfiles most of the time.
1085 There is the place for the Z loop!
1087 But the combination is something to collect in a database, isn't it. Did
1088 csync2 just harrumph?
1090 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1092 * it just occurred to me that hosts in the same mirroring pool could
1093 help out each other even without rewriting the recentfile. Just fetch
1094 the stuff to mirror from several places, bingo. But that's something
1095 that should rather live in a separate package or in rsync directly.
1097 * cronjobs are unsuited because with ntp they would all come at the full
1098 minute and disturb each other. Besides that I'd hate to have a backbone
1099 with more than a few seconds latency.
1101 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1103 * a second rsync server with access control for PAUSE. Port? 873 is the
1104 standard port, let's take 8873.
1106 * if there were a filesystem based on this, it would have a slow access
1107 to inexistent files. It would probably provide wrong readdir (only based
1108 on current content) or also a slow one (based on a recentfile written
1109 after the call). But it would provide fast access to existing files. Or
1110 one would deliberately allow slightly blurred answers based on some
1111 sqlite reflection of the recentfiles.
1113 * todo: write a variant of mirror() that combines two or more
1114 recentfiles and treats them like one
1116 * todo: signal handler to remove the tempfile
1118 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1120 * now that we have the symlink I forgot how it should be used in
1123 * the z loop: add missing files to Z file. Just append them (instead of
1124 prepending). So one guy prepends something from the Y file from time to
1125 time and another guy appends something rather frequently. Collecting
1126 pond. When Y merges into Z, things get epoch and the collecting pond
1127 gets smaller. What exactly are "missing files"?
1129 take note of current epoch of the alpha file, let's call it the
1132 find all files on disk
1134 remove all files registered in the recentworld up to recent-ts
1136 remove all files that have been deleted after recent-ts according to
1139 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1141 * rersyncrecent might be a cronjob with a (locked) state file which
1142 contains things like after and maybe last z sync or such?
1144 rrr-mirror might be an alternative name but how would we justify the
1145 three Rs when there is no Re-Rsync-Recent?
1147 With the --loop parameter it is an endless loop, without it is no loop.
1148 At least this is simple.
1150 * todo: new accssor z-interval specifies how often the Z file is updated
1151 against the filesystem. We probably want no epoch stamp on these
1152 entries. And we want to be able to filter the entries (e.g. no
1153 by-modules and by-category tree)
1155 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1157 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1158 into Z. We do not need the other files filled up with filesystem
1161 * need interface to query for a file in order to NOT call update on
1162 PAUSE a second time within a short time.
1164 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1166 * recommended update interval? Makes no sense, is different for
1173 change-log-default-name: "Todo"