| blob | 0c8d2aeb34511e185867bea50b5439db661b5bf8 |
1 2009-04-13 Andreas J. Koenig <andk@cpan.org>
3 * Study http://en.wikipedia.org/wiki/Magnet_URI_scheme
5 * is it true (as stated at
6 https://fedorahosted.org/InstantMirror/wiki/ExistingRepositoryReplicationMethods)
7 that rsync can lead to errors when upstream changes before client sync
8 has completed? Can we see the error when we run this on upstream:
10 % perl -e '
11 use Time::HiRes qw(time);
12 while (){
13 open my $fh, ">", "changingfile.txt.new" or die;
14 print $fh (time."\n") x 1000000;
15 close $fh;
16 rename "changingfile.txt.new", "changingfile.txt" or die;
17 }
18 '
20 And on the receiving end:
22 % while true; do
23 rsync k75:`pwd`/changingfile.txt .; cat changingfile.txt| uniq|wc -l
24 sleep 1;
25 done
27 I cannot get it to fail. Apparently it is sufficient when upstream
28 always writes atomically (which of course is mandatory)?
30 * from ABH:
32 > https://fedorahosted.org/InstantMirror/
33 > https://www.redhat.com/mailman/listinfo/instantmirror-list
34 > irc.freenode.net channel #instantmirror
36 A cool name for a project. Inspires me to write a few sentences about
37 bittorrent's role in the grand picture.
39 http://spreadsheets.google.com/pub?key=pGlWX10blP4u2kM05SDtiMg is a
40 spreadsheet collected by Atul Aggarwal about bittorrent implementations.
42 2009-04-12 Andreas J. Koenig <andk@cpan.org>
44 * Interesting last minute bug during real download testing: the output
45 isn't as pretty anymore, it seems that more work is being done than
46 needed.
48 Introducing a highlevel _alluptodate method helped a lot in debugging
49 this.
51 FIXED now. Reason was that dirtymark on PAUSE currently is broken; it's
52 different on 1h file than on the other files. This lead to frequent
53 calls to _fullseed. Limiting the dirtymark check to $i==0 resolved the
54 issue.
56 2009-04-12 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
58 * Todo: try if we can be faster with a native float package. I'm really
59 glad to have a machine exposing perl floating point bugs, it forced me
60 to retract this item from the changes file: "several speedups in the
61 fakebigfloat code" before I get cpan testers fails.
63 2009-04-11 Andreas J. Koenig <andk@cpan.org>
65 * Bug: when the dirtymark changes upstream, then the downstream server
66 notices it immediately and mirrors the "1h" file but then leaves the
67 rmirror loop and continues with "6h" on the next round through. This
68 behaviour goes on until it reaches the Z file. Because we eat the Z file
69 piecemeal we leave the loop after a while but after that everybody seems
70 to have forgotten that there is some work left to be done. Bad, bad,
71 bad.
73 ETOOCONFUSING better name?
75 Recentfile::get_remote_recentfile_as_tempfile OK
76 Recentfile::resolve_recentfilename split_rfilename DONE
78 Recent::_principal_recentfile_object _principal_recentfile_fromremote
79 Recent::principal_recentfile OK
80 Recent::_resolve_rfilename _principal_recentfile_fromremote_resosymlink
82 some void, some not, some not void but called in void context.
84 2009-04-10 Andreas J. Koenig <andk@cpan.org>
86 * Todo: rename rrr-dirtyupdate to rrr-update and add an option --dirty
87 so we can use it for normal update with a single file.
89 * Bug? When the dirtymark changes then the second tier hosts quickly
90 reset their DONE state and restart mirroring. But then they mirror
91 potentially outdated index files with inconsistent dirtymark because the
92 first tier box may be a bit behind on the large recentfiles.
94 This means we need a brake when iterating over the recentfiles that
95 refuses to use a recentfile with the wrong dirtymark.
97 2009-04-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
99 * install the bin/* files and talk about rrr-overview somewhere.
101 * Ask B. Hansen asks for tempdir accessor such that tempfiles get
102 created outside the target tree.
104 * Barbie asks for better logging capabilities
106 2009-03-29 Andreas J. Koenig <andk@cpan.org>
108 * known bug: cannot configure the Recentfile objects to keep delete
109 events. How to call the accessor for that? keep_delete_objects_forever?
111 Sounds acceptable. FIXED (but untested).
113 * known bug: keeps temporary index files lying around.
115 2009-03-24 Andreas J. Koenig <andk@cpan.org>
117 * Some equivalent for
119 for ( @{$rrr->recentfiles} ) { $_->verbose(0) }
121 ? DONE
123 2009-03-22 Andreas J. Koenig <andk@cpan.org>
125 * k81 is again client of k75 with different parameters.
127 * broken now: (1) seeding and unseeding: rmirror is seeding and talking
128 about it all the time and nobody reacts accordingly; (2) lots of temp
129 files get created and not removed; culprit the new call to
130 get_remote_recentfile_as_tempfile within Recent.pm; the manpage says the
131 caller has to remove the tempfile after use.
133 Need the drawing board.
135 PARTLY FIXED: retracting the idea to call
136 get_remote_recentfile_as_tempfile from Recent.pm and moving around seed
137 and unseed such that uptodate now means "we have mirrored this
138 recentfile's bunch of files" and seeded means "somebody believes the
139 index file for this rf needs be refreshed". This seems to work now.
141 Still collecting temp files that nobody cares to depollute.
143 2009-03-21 Andreas J. Koenig <andk@cpan.org>
145 * Bug?: should it be harder than it is atm to set the timestamp to the
146 future? YES, FIXED
148 * bug with native integers:
150 -
151 epoch: 997872011
152 path: id/D/DE/DELTA/Crypt-Rijndael_PP-0.03.readme
153 type: new
154 -
155 epoch: 1195248431
156 path: id/L/LG/LGODDARD/Tk-Wizard-2.124.readme
157 type: new
159 Native integer broke when native math was turned off. FIXED
161 * Bug: something between id/P/PH/PHISH/CGI-XMLApplication-1.1.2.readme
162 and id/C/CH/CHOGAN/HTML-WWWTheme-1.06.readme. Mirroring the Z file loops
163 in this area.
165 Yes, records out of order:
167 447003 -
168 447004 epoch: 995885533
169 447005 path: id/P/PH/PHISH/CGI-XMLApplication_0.9.3.readme
170 447006 type: new
171 447007 -
172 447008 epoch: 995890358
173 447009 path: id/H/HD/HDIAS/Mail-Cclient-1.3.readme
174 447010 type: new
175 447011 -
176 447012 epoch: 995892221
177 447013 path: id/H/HD/HDIAS/Mail-Cclient-1.3.tar.gz
178 447014 type: new
180 FIXED with sanity check and later with the integer fix.
182 * Bug: want the index files in a .recent directory
184 * Bug: lots of dot files are not deleted in time
186 * possible test case: can a delete change the timestamp? This would
187 probably break the order of events.
189 2009-03-20 Andreas J. Koenig <andk@cpan.org>
191 * 1233701831.34486 what's so special about this number/string? It
192 belongs to
193 id/G/GR/GRODITI/MooseX-Emulate-Class-Accessor-Fast-0.00800.tar.gz and
194 atm lives in Y,Q, and Z.
196 It is the first entry after id/--skip-locking which has timestamp
197 1234164228.11325 which represents a file that doesn't exist anymore.
199 * another bug:
201 Sync 1237531537 (31547/33111/Z) id/J/JH/JHI/String-Approx-2.7.tar.gz ...
202 _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
203 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
204 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
205 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
206 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
207 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
208 at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 76
209 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 131
210 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatlt(1237505213.21133, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 110
211 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
212 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
213 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
215 FIXED, it was the "--skip-locking" file where manual intervention was participating
217 * bug on the mirroring slave: when the dirtymark gets increased we
218 probably do not reset the done intervals. The mirrorer stays within
219 tight bounds where it tries to sync with upstream and never seems to
220 finish. In the debugging state file I see lots of identical intervals
221 that do not get collapsed. When I restart the mirrorer it dies with:
223 Sync 1237507989 (227/33111/Z) id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz ...
224 _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
225 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatcmp(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/FakeBigFloat.pm line 101
226 File::Rsync::Mirror::Recentfile::FakeBigFloat::_bigfloatge(1237400817.94363, undef) called at /home/k/sources/rersyncrecent/lib/File/Rsync/Mirror/Recentfile/Done.pm line 226
227 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
228 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
229 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
230 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
231 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
232 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
234 and debugging stands at
236 At: before
237 Brfinterval: Z
238 Ci: 227
239 Dre+1:
240 epoch: 1237400802.5789
241 path: id/J/JO/JOHND/CHECKSUMS
242 type: new
243 Dre-0:
244 epoch: 1237400807.97514
245 path: id/M/MI/MIYAGAWA/CHECKSUMS
246 type: new
247 Dre-1:
248 epoch: 1237400817.94363
249 path: id/X/XI/XINMING/Catalyst-Plugin-Compress.tar.gz
250 type: new
251 Eintervals:
252 -
253 - 900644040
254 - 900644040
255 - []
257 and it is reproducable.
259 Why does the mirrorer not fetch a newer Z file? It is 18 hours old while
260 pause has a fresh one.
262 FIXED, it was the third anded term in each of the ifs in the IV block in
263 _register_one: with that we make sure that we do not stamp on valuable
264 interval data.
266 2009-03-17 Andreas J. Koenig <andk@cpan.org>
268 * done: verified the existence of the floating point bug in bleadperl
269 and verified that switching from YAML::Syck to YAML::XS does not resolve
270 it.
272 BTW, the switch was doable with
274 perl -i~ -pe 's/Syck/XS/g' lib/**/*.pm t/*.t
276 and should be considered as a separate TODO
278 * todo: integrate a dirty update with two aggregate calls before
279 unlocking for frictionless dirtying
281 * todo: start the second rsync daemon on pause
283 * todo: move index files to .recent: this cannot simply be done by
284 setting filenameroot to .recent/RECENT. Other parts of the modules rely
285 on the fact that dirname(recentfile) is the root of the mirrored tree.
287 2009-03-16 Andreas J. Koenig <andk@cpan.org>
289 * What was the resolution of the mirror.pl delete hook bug? Do we call
290 the delete hook when pause removes a file from MUIR?
292 * Today on pause: Updating 2a13fba..29f284d and installing it for
293 /usr/local/perl-5.10.0{,-RC2}
295 TURUGINA/Set-Intersection-0.01.tar.gz was the last upload before this
296 action and G/GW/GWILLIAMS/RDF-Query-2.100_01.tar.gz the first after it
298 2009-03-15 Andreas J. Koenig <andk@cpan.org>
300 * currently recent_events has the side effect of setting dirtymark
301 because it forces a read on the file. That should be transparent, so
302 that the dirtymark call always forces a cache-able(?) read.
304 * The bug below is -- after a lot of trying -- not reproducible on a
305 small script, only in the large test script. The closest to the output
306 below was:
308 #!perl
309 use strict;
310 use Devel::Peek;
311 my $x = "01237123229.8814";
312 my($l,$r);
313 for ($l,$r) {
314 $_ = "x"x34;
315 }
316 ($l,$r) = ($1,$2) if $x =~ /(.)(.+)/;
317 $r = int $r;
318 $l = "1237123231.22458";
319 $r = "1237123231.22458";
320 1 if $l/1.1;
321 Devel::Peek::Dump $l;
322 Devel::Peek::Dump $r;
323 Devel::Peek::Dump $x = $l <=> $r;
324 die "BROKE" if $x;
325 __END__
327 The checked in state at c404a85 fails the test with my
328 /usr/local/perl-5.10-uld/bin/perl on 64bit but curiously not with
329 /usr/local/perl-5.10-g/bin/perl. So it seems the behaviour is not even
330 in the test script always consistent.
332 * Todo: write a test that inserts a second dirty file with an already
333 existing timestamp. DONE
335 * Bug in perl 5.10 on my 64bit box:
337 DB<98> Devel::Peek::Dump $l
338 SV = PVMG(0x19e0450) at 0x142a550
339 REFCNT = 2
340 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
341 IV = 0
342 NV = 1237123231.22458
343 PV = 0x194ce70 "1237123231.22458"\0
344 CUR = 16
345 LEN = 40
347 DB<99> Devel::Peek::Dump $r
348 SV = PVMG(0x19e0240) at 0x142a3e8
349 REFCNT = 2
350 FLAGS = (PADMY,POK,pPOK)
351 IV = 1237123229
352 NV = 1237123229.8814
353 PV = 0x19ff900 "1237123231.22458"\0
354 CUR = 16
355 LEN = 40
357 DB<100> Devel::Peek::Dump $l <=> $r
358 SV = IV(0x19ea6e8) at 0x19ea6f0
359 REFCNT = 1
360 FLAGS = (PADTMP,IOK,pIOK)
361 IV = -1
363 DB<101> Devel::Peek::Dump $l
364 SV = PVMG(0x19e0450) at 0x142a550
365 REFCNT = 2
366 FLAGS = (PADMY,NOK,POK,pNOK,pPOK)
367 IV = 0
368 NV = 1237123231.22458
369 PV = 0x194ce70 "1237123231.22458"\0
370 CUR = 16
371 LEN = 40
373 DB<102> Devel::Peek::Dump $r
374 SV = PVMG(0x19e0240) at 0x142a3e8
375 REFCNT = 2
376 FLAGS = (PADMY,NOK,POK,pIOK,pNOK,pPOK)
377 IV = 1237123231
378 NV = 1237123231.22458
379 PV = 0x19ff900 "1237123231.22458"\0
380 CUR = 16
381 LEN = 40
383 Retry with uselongdouble gives same effect. Not reproducable on 32bit box (k75).
385 * Todo: reset "done" or "covered" and "minmax" after a dirty operation?
387 2009-03-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
389 * $obj->merge ($other) needs to learn about equal epoch which may happen
390 since dirty_epoch intruded.
392 * Wontfix anytime soon: I think we currently do not support mkdir. Only
393 files!
395 2009-01-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
397 * Todo: continue working on update(...,$dirty_epoch). It must be
398 followed by a fast_aggregate!
400 2008-12-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
402 * maybe we need a closest_entry or fitting_interval or something like
403 that. We want to merge an event into the middle of some recentfile.
404 First we do not know which file, then we do not know where to lock,
405 where to enter the new item, when and where to correct the dirtymark.
407 So my thought is we should first find which file.
409 Another part of my brain answers: what would happen if we would enter
410 the new file into the smallest file just like an ordinary new event,
411 just as an old event?
413 (1) we would write a duplicate timestamp? No, this would be easy to
414 avoid
416 (2) we would make the file large quickly? Yes, but so what? We are
417 changing the dirtymark, so are willing to disturb the downstream hosts.
419 2008-11-22 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
421 * 10705 root 17 0 725m 710m 1712 S 0.0 46.8 834:56.05 /home/src/perl/repoperls/installed-perls/perl/pVNtS9N/perl-5.8.0@32642/bin/perl -Ilib /home/k/sources/CPAN/GIT/trunk/bin/testing-rmirror.pl
423 leak!
425 https://rt.cpan.org/Ticket/Display.html?id=41199
427 * bzcat uploads.csv.bz2 | perl -F, -nale '$Seen{$F[-1]}++ and print'
429 Strangest output being HAKANARDO who managed to upload
431 Here is a better oneliner that includes also the first line of each
432 finding:
434 bzcat uploads.csv.bz2 | perl -MYAML::Syck -F, -nale '$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)}'
436 2008-10-31 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
438 * memory leak in the syncher? It currently weighs 100M.
440 Update 2008-11-02:
442 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
445 2008-10-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
447 * lookup by epoch and by path and use this ability on the pause to never
448 again register a file twice that doesn't need it. Let's call it
449 contains().
451 * after the dirtymark is done: fill up recentfiles with fake (historic)
452 entries; fill up with individual corrections; algorithm maybe to be done
453 with bigfloat so that we can always place something in the middle
454 between two entries. Before we must switch to bigfloat we could try to
455 use Data::Float::nextup to get the.
457 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
458 directly.
460 * dirtymark now lives in Recentfile, needs to be used in rmirror.
462 * find out why the downloader died after a couple of hours without a net
463 connection. Write a test that survives the not-existence of the other
464 end forever.
466 2008-10-15 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
468 * reconsider the HTTP epoch only. Not the whole thing over HTTP because
469 it makes less sense with tight coupling for secondary files. But asking
470 the server what the current epoch is might be cheaper on HTTP than on
471 rsync. (Needs to be evaluated)
473 * remove the 0.00 from the verbose overview in the Merged column in the
474 Z row. DONE
476 * write tests that expose the problems of the last few days: cascading
477 client/server roles, tight coupling for secondary RFs, deletes after
478 copies.
480 * Some day we might want to have policy options for the slave:
481 tight/loose/no coupling with upstream for secondary RFs. tight is what
482 we have now. loose would wait until a gap occurs that can be closed.
484 2008-10-14 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
486 * revisit all $rfs->[$i+1] places if they now make sense still
488 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
490 * another bug is the fact that the mirror command deletes files before
491 it unhides the index file, thus confusing downstream slaves. We must not
492 delete before unhiding and must delete after unhiding. FIXED.
494 * new complication about the slave that is playing a server role.
495 Currently we mirror from newest to oldest with a hidden temporary file
496 as index. And when one file is finished, we unhide the index file.
497 Imagine the cascading server/slave is dead for a day. It then starts
498 mirroring again with the freshest thing and unhides the freshest index
499 file when it has worked through it. In that moment it exposes a time
500 hole. Because it now works on the second recentfile which is still
501 hidden.
503 We currently do nothing special to converge after such a drop out. At
504 least not intentionally and robustly and thought through.
506 The algorithm we use to seed the next file needs quite a lot of more
507 robustness than it currently has. Something to do with looking at the
508 merged element of the next rf and when it has dropped off, we seed
509 immediately. And if it ramains dropped off, we seed again, of course.
511 Nope, looking from smaller to larger RFS we look at the merged element
512 of this RF and at the minmax/max element of the next RF. If that
513 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
514 seeding it.
516 And we need a public accessor seed and unseed or seeded. But not the mix
517 of public and private stuff that then is used behind the back.
519 And then the secondary* stuff must go.
521 And we must understand what the impact is on the DONE system. Can it go
522 unnoticed that there was a hole? And could the DONE system have decided
523 the hole is covered? This should be testable with three directories where
524 the middle stops working for a while. Done->merge is suspicious, we must
525 stop it from merging non-conflatable neighbors due to broken continuity.
527 FIXED
529 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
531 * Slaven suggests to have the current epoch or the whole current
532 recentfile available from the HTTP server and take it away with
533 keepalive. This direction goes the granularity down to subseconds.
535 We might want to rewrite everything to factor out transport and allow
536 the whole thing to run via HTTP.
538 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
540 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
541 symlink?
543 * smoker on k81 fetching from k75 to verify cascading works. See
544 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
546 * maybe the loop should wait for CHECKSUMS file after every upload. And
547 CPAN.pm needs to deal with timestamps in the future.
549 * do not forget the dirtymark!
551 Text: have a new flag on recentfiles with the meaning: if this
552 changes, you're required to run a full rsync over all the files. The
553 reason why we set it would probably be: some foul happened. we injected
554 files in arbitrary places or didn't inject them although they changed.
555 The content of the flag? Timestamp? The relation between the
556 recentfiles would have to be inheritance from the principal, because any
557 out of band changes would soon later propagate to the next recentfile.
559 By upping the flag often one can easily ruin the slaves.
561 last out of band change? dirtymark?
563 Anyway, this implies that we read a potentially existing recentfile
564 before we write one.
566 And it implies that we have an eventloop that keeps us busy in 2-3
567 cycles, one for current stuff (tight loop) and one for the recentfiles
568 (cascade when principal has changed), one for the old stuff after a
569 dirtymark change.
571 And it implies that the out-of-band change in any of the recentfiles
572 must have a lock on the principal file and there is the place to set the
573 dirtymark.
575 * start a FAQ, especially quick start guide questions. Also to aid those
576 problematic areas where we have no good solution, like the "links"
577 option to rsync.
579 * wish feedback when we are slow.
581 * reduce mccabe
583 * Remove a few DEBUG statements.
585 * The multiple-rrr way of doing things needs a new option to rmirror,
586 like piecemeal or so. Not urgent because after the first pass through,
587 things run smoothely. It's only ugly during the first pass.
589 * I have the suspicion that the code is broken that decides if the
590 neighboring RF needs to be seeded. I fear when too much time has gone
591 between two calls (in our case more than one hour), it would not seed
592 the neighbor. Of course this will never be noticed, so we need a good
593 test for it.
595 * local/localroot confusion: I currently pass both options but one must
596 do.
598 * accounts for early birds on PAUSE rsync daemon.
600 * hardcoded 20 seconds
602 * who mirrors the index? DOING now.
604 * which CPAN mirrors offer rsync?
606 * visit all XXX, visit all _float places
608 * rename the pathdb stuff, it's too confusing. No idea how.
610 * rrr-inotify, backpan, rrr-register
612 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
614 * current bugs: the pathdb seems to get no reset, the seeding of the
615 secondaryttl stuff seems not to have an effect. Have helped myself with
616 a rand(10), need to fix this back. So not checked in. Does the rand
617 thing even help?
619 The rand thing helps. The secondaryttl stuff was in the wrong line,
620 fixed now.
622 The pathdb stuff was because I called either _pathdb or __pathdb on the
623 wrong object. FIXED now.
625 * It's not so beautiful if we never fetch the recentfiles that are not
626 the principal, even if this is correct behaviour. We really do not need
627 them after we have fetched the whole content.
629 OK, we want a switch for that: secondaryttl DONE
631 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
633 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
634 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
635 it lets 40 follow. FIXED
637 * See that the long running process really only updates the principal
638 file unless it has missed a timespan during which something happened. If
639 nothing happened, it must notice even when it misses the timespan. DONE
641 * we must throw away the pathdb when we have reached the end of Z. From
642 that moment we can have a very small pathdb because the only reason for
643 a pathdb is that we know to ignore old records in old files. We won't
644 need this pathdb again before the next full pass over the data is
645 necessary and then we will rebuild it as we go along. DONE
647 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
649 * I think, Done::register_one is doing wrong in that it does not
650 conflate neighboring pieces. The covered() method cannot do this because
651 it has no recent_events array at hand. But register_one has it and could
652 do it and for some reason misses to do it (sometimes).
654 This means that the three tests I just wrote can probably not survive
655 because they test with an already broken Done structure.
657 The art now is to detect how it happens, then to reproduce, then write a
658 test, then fix it.
660 So from the logfile this is what happens: we have a good interval with
661 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
662 top of it. Locally we now mirror F2 and open a new done interval for it.
663 Then we mirror F1 but this time with the timestamp T1b. And when we then
664 try to close the gap, we do not find T1 but instead something older. We
665 should gladly accept this older piece and this would fix this bug.
667 FIXED
669 * bug to fix: when the 1h file changes while rmirror is running, we do
670 correctly sync the new files but never switch to the 6h file but rather
671 stay in a rather quick loop that fetches the 1h file again and again.
673 Is it possible that we initialize a new object? Or does
674 get_remote_recentfile_as_tempfile overwrite something in myself?
676 Want a new option: _runstatusfile => $file which frequently dumps the
677 state of all recentfiles to a file.
679 FIXED
681 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
683 * Todo: now teach update to verify the timestamp is about to write
684 against the previous and use _increase_a_bit if it doesn't comply with
685 strict monotony. DONE
687 * The problem of rounding. So far perl's default precision was
688 sufficient. One day it won't be. FakeFloat has an easy job when it is
689 only reading and other machines have written correctly. But when we want
690 to write a floating point number that is a bit larger than the other
691 one, then we need our own idea of precision.
693 Slaven said: just append a "1". This might be going towards the end of
694 usability too quickly. I'd like something that actually uses the decimal
695 system. Well, appending a 1 also does this but...
697 E.g. we have 1.0. nextup on this architecture is starting with
698 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
699 taken the 1.0000000000000003 and the next user comes and the time tells
700 him 1.0 again. He has to beat my number without stepping over the
701 nextup. This is much less space than I had when I chose 1,2,3.
703 What is also irritating is that nextup is architecture dependent. The
704 128 bit guy must choose very long numbers to fit in between whereas the
705 other one with 16 bit uses larger steps. But then the algorithm is the
706 same for both, so that would be a nice thing.
708 I see two situation where we need this. One is when Time::HiRes returns
709 us a value that is <= the last entry in our recentfile. In this case
710 (let's call it the end-case) we must fill the region between that number
711 and the next higher native floating point number. The other is when we
712 inject an old file into an old recentfile (we would then also set a new
713 dirtymark). We find the integer value already taken and need a slightly
714 different one (let's call it the middle-case). The difference between
715 the two situations is that the next user will want to find something
716 higher than my number in the end-case and something lower than my number
717 in the middle case.
719 So I suggest we give the function both a value and an upper bound and it
720 calculates us a primitive middle. The upper bound in the middle-case is
721 the next integer. The upper bound on the end-case is the nextup floating
722 point number. But the latter poses another problem: if we have occupied
723 the middle m between x and nextup(x), then the nextup(m) will probably
724 not be the same as nextup(x) because some rounding will take place
725 before the nextup is calculated and when the rounding reaches the
726 nextup(x), we will end up at nextup(nextup(x)).
728 So we really need to consider the nextup and the nextdown from there and
729 then the middle and that's the number we may approach asymptotically.
730 Ugly. But DONE.
732 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
734 * consider deprecating the use of RECENT.recent as a symlink. It turns
735 out to need extra hoops with the rsync options and just isn't worth it.
736 Or maybe these extra hoops are needed anyway for the rest of the tree?
737 Nope, can't be the case because not all filesystems support symlinks.
739 But before doing the large step, I'll deprecate the call of
740 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
741 only introduced to resolve RECENT.recent and complicates the routine far
742 beyond what it deserves.
744 DONE. Won't deprecate RECENT.recent, just moved its handling to the
745 supervisor.
747 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
749 * I think it's a bug that the rsync_option links must be set to true in
750 order to support RECENT.recent and that nobody cares to set it
751 automatically. Similar for ignore_link_stat_errors.
753 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
755 * Todo: find all todos together and make a plan what is missing for a
756 release.
758 - verifytree or something like that. fsck maybe.
760 - rersyncrecent, the script itself? What it do?
762 - a way to only mirror the recentfiles without mirroring the whole
763 remote system such that people can decide to mirror only partially see
764 also 2008-08-30. .shadow-xxx directory? this also needed for a
765 filesystem that is still incomplete and might need the mirrorfiles for
766 lookup(?)
768 - long living objects that mirror again and again. Inject something
769 into ta, see how it goes over to tb.
771 - how do we continue filling up the DONE system when we use an object
772 for the second time? "fully covered" and "uptodate" or new terminology.
774 - overview called on the wrong file should be understandable
776 - the meta data field that must change when we fake something up so that
777 the downstream people know they have to re-fetch everything.
779 - how tolerant are we against missing files upstream? how do we keep
780 track? there are legitimate cases where we did read upstream index right
781 before a file got deleted there and then find that file as new and want
782 it. There are other cases that are not self healing and must be tracked
783 and bugreported.
785 - how, exactly, do we have to deal with deletes? With rsync errors?
787 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
788 authors) failed: No such file or directory (2)
790 The file above is a delete in 1h and a new in file 1M and the
791 delete in the locally running rmirror did not get propagated to the 1M
792 object. Bug. And the consequence is a standstill.
794 It seems that a slave that works with a file below the principal needs
795 to merge things all the way up to get rid of later deletes. Or keep
796 track of all deletes and skip them later. So we need a trackdeletes.pm
797 similar to the done.pm?
799 see also 2008-08-20 about spurious deletes that really have no add
800 counterpart and yet they are not wrong.
802 - consider the effect when resyncing the recentfile takes longer than
803 the time per loop. Then we never rsync any file. We need to diagnose
804 that and force an increase of that loop time. But when we later are fast
805 enough again because the net has recovered, then we need to switch back
806 to original parameters. ERm, no, it's enough to keep syncing at least
807 one file before refetching an index file.
809 - remember to verify that no temp files are left lying around and the
810 signal handler
812 - status file for not long running jobs that want to track upstream with
813 a, say, cronjob.
815 - revisit all XXX _float areas and study Sub::Exporter DONE
817 - persistent DB even though we just said we do not need it. Just for
818 extended capabilities and time savings when, for example, upstream
819 announces a reset and we get new recentfiles and could then limit
820 ourselves to a subset of files (those that have a changed epoch) in a
821 first pass and would only then do the loop to verify the rest. Or
822 something.
824 * Todo: aggregate files should know their feed and finding the principal
825 should be done stepwise. (?)
827 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
828 left locks around. DONE
830 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
832 * maybe extend the _overview so that it always says if and where the
833 last file is in the next file and where the next event in the next rf
834 would lie. No, don't like this anymore. REJECT
836 * take the two new redundant tests out again, only the third must
837 survive. DONE
839 * Todo: add a sanity check if the merged structure is really pointing to
840 a different rf and that this different rf is larger. DONE
842 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
844 * now test, if they are overlapping. And test if there is a file in the
845 next rf that would fit into this rf's interval.
847 1h 1222324012.8474 1222322541.7963 0.4086
848 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
849 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
850 1W 1222313218.3626 1221708477.5829 167.9835
852 I suspect that somebody writes a merged timestamp without having merged
853 and then somebody else relies on it.
855 If aggregate is running, the intervals must not be extravagated, if it
856 is not running, there must not be bounds, the total number of events in
857 the system must be counted and must be controlled throughout the tests.
858 That the test required the additional update was probably nonsense,
859 because aggregate can cut pieces too. FIXED & DONE
861 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
863 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
864 changed. FIXED
866 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
868 * Most apparent bug at the moment is that the recentfiles are fetched
869 too often. Only the principal should be fetched and if it has not
870 changed, the others should not be refetched. ATM I must admit that I'm
871 happy that we refetch more often than needed because I can more easily
872 fix bugs while the thing is running.
874 * Let's say, 1220474966.19501 is a timestamp of a file that is already
875 done but the done system does not know about it. The reason for the
876 failure is not known and we never reach the status uptodate because of
877 this. We must get over it.
879 Later it turns out that the origin server had a bug somewhere.
880 1220474966.19042 came after 1220474966.19501. Or better: it was in the
881 array of the recentfile one position above. The bug was my own.
883 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
885 * There is the race condition where the server does a delete and the
886 slave does not yet know and then tries to download it because he sees
887 the new. So for this time window we must be more tolerant against
888 failure. If we cannot download a file, we should just skip it and should
889 not retry immediately. The whole system should discover the lost thing
890 later. Keeping track with the DONE system should really be a no brainer.
892 But there is something more: the whole filesystem is a database and the
893 recentfiles are one possible representation of it. It's a pretty useful
894 representation I think that's why I have implemented something around
895 it. But for strictly local operation it has little value. For local
896 operation we would much rather have a database. So we would enter every
897 recentfile reading and every rsync operation and for every file the last
898 state change and what it leads to. Then we would always ignore older
899 records without the efforts involved with recentfiles.
901 The database would have: path,recentepoch,rsyncedon,deletedon
903 Oh well, not yet clear where this leads to.
905 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
907 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
908 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
910 * statusfile, maybe only on demand, alone to have a sharp debugging
911 tool. It is locked and all recentfiles dump themselves into it and we
912 can build a viewer that lets us know where we stand and what's inside.
914 * remember: only the principal recentfile needs expiration, all others
915 shall be expired by principal if it discovers that something has move
916 upstream.
918 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
920 * Always check if we stringify to a higher value than in the entry
921 before. DONE
923 * And in covered make an additional check if we would be able to see a
924 numerical difference between the two numbers and if we can't then switch
925 to a different, more expensive algorithm. Do not want to be caught by
926 floating surprises. DONE
928 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
930 * caching has several aspects here: we can cache the interval of the
931 recentfile which only will change when the mtime of the file changes. We
932 must re-mirror the recentfile when its ttl has expired. Does have_read
933 tell you anything? It counts nothing at all. Only the mtime is
934 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
935 thing the have_mirrored because it is unrelated to the mtime.
937 * Robustness of floating point calculations! I always thought that the
938 string calculated by the origin server for the floating representation
939 of the epoch time is just a string. When we convert it to a number and
940 later back to a string, the other computer might come to a different
941 conclusion. This must not happen, we want to preserve it under any
942 circumstances. I will have to write tests with overlong sequences that
943 get lost in arithmetic and must see if all still works well. DONE
945 But one fragile point remains: if one host considers a>b and the other
946 one considers them == but no eq. To prevent this, we must probably do
947 some extra homework. DONE
949 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
951 * the concept of tracking DONE needs an object per recentfile that has
952 something like these methods:
954 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
956 covered() register() covered()
958 The unclear thing is how we translate points in time into intervals. We
959 could pass a reference to the current recent_events array when running
960 we_have(xxx) and let the DONE object iterate over it such that it only
961 has to store a list of intervals that can melt into each other. Ah, even
962 passing the list together with a list of indexes seems feasiable.
964 Or maybe ask for the inverted list?
966 Whenever the complete array is covered by the interval we say we are
967 fully covered and if the recentfile is not expired, we are uptodate.
969 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
971 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
973 * need a way to "return" the next entry after the end of a list. When
974 the caller says "before" or "after" we would like to know if he could
975 cover that interval/threshold or not because this influences the effect
976 of a newer timestamp of that recentfile. DONE with $opt{info}.
978 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
980 * one of the next things to tackle: the equivalent of csync2 -TIXU.
982 loop implies tixu (?). Nope, something like --statefile decides. Per
983 default we do ...?
985 T test, I init, X including removals, U nodirtymark
987 So we have no concept of dirtymarks, we only trust that since we are
988 running we have observed everything steadily. But people will not let
989 this program run forever so we must consider both startup penalty and
990 book keeping for later runs. We keep this for later. For now we write a
991 long running mirror that merges several intervals.
993 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
995 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
996 the intervals!
998 * rersyncrecent, the script: default to one week. The name of the switch
999 is --after. Other switches? --loop!
1001 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1003 * need a switch --skip-deletes (?)
1005 * need a switch --enduser that tells us that the whole tempfile
1006 discipline is not needed when there is no downstream user. (?)
1008 Without this switch we cannot have a reasonable recent.pl that just
1009 displays the recent additions. Either we accept to download everything.
1010 Or we download temporary files without the typical rsync protocol
1011 advantages.
1013 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
1014 File::Temp::tempdir, this might be a win.
1016 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1018 * apropos missing: we have no push, we never know the downstream
1019 servers. People who know their downstream hosts and want to ascertain
1020 something will want additional methods we have never thought about, like
1021 update or delete a certain file.
1023 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1025 * tempted to refactor rmirror into resolve_symlink, localize, etc.
1026 Curious if rsync_options=links equal 0 vs. 1 will make the expected
1027 difference.
1029 * rsync options: it's a bit of a pain that we usually need several rsync
1030 options, like compress, links, times, checksum and that there is no
1031 reasonable default except the original rsync default. I think wee can
1032 safely assume that the rsync options are shared between all recentfile
1033 instances within one recent tree.
1035 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1037 * deletes: if a delete follows an add quickly enough it may happen that
1038 a downstream mirror did not see the add at all! It seems this needs to
1039 be mentioned somewhere. The point here is that even if the downstream is
1040 never missing the principal timeframe it may encounter a "delete" that
1041 has no complimentary "add" anywhere.
1043 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1045 * I suspect the treat of metadata is incorrect during read or something.
1046 The bug that I am watching is that between 06:08 and 06:09 the 6h file
1047 contained more than 6 hours worth of data. At 06:08 we merged into the
1048 1d file. We need to take snapshots of the 6h file over the course of an
1049 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
1050 enough.
1052 Much worse: watching the 1h file: right at the moment (at 06:35) it
1053 covers 1218867584-1219120397 which is 70 hours.
1055 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
1056 08:19:44 2008, that is when I checked out last time, so it seems to be
1057 aggregating and never truncating?
1059 No, correct is: it is never truncating; but wrong is: it is aggregating.
1060 It does receive a lot of events from time to time from a larger file.
1061 Somehow a large file gets merged into the small one and because the
1062 "meta/merged" attribute is missing, nobody is paying attention. I
1063 believe that I can fix this by making sure that metadata are honoured
1064 during read. DONE and test adjusted.
1066 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1068 * grand renaming plan
1070 remotebase => remoteroot to fit well with localroot DONE
1071 local_path() => localroot seems to me should already work DONE
1072 recentfile_basename => rfilename no need to stress it has no slash DONE
1074 filenameroot??? Doesn't seem too bad to me today. Maybe something like
1075 kern? It would anyway need a deprecation cycle because it is an
1076 important constructor.
1078 * I like the portability that Data::Serializer brings us but the price
1079 is that some day we might find out that it is slowing us a bit. We'll
1080 see.
1082 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1084 * should we not enter the interval of the principal (or the interval of
1085 the merging file?) in every aggregated/merged file?
1087 * we should aim at a first release and give up on thinking about
1088 sanitizing stuff and zloop. Let's just admit that a full traditional
1089 rsync is the only available sanitizer ATM. Otherwise it's complicated
1090 stuff: sanitizing on the origin server, sanitizing on the slaves,
1091 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
1092 get the basics out before this becomes a major cause for mess.
1094 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
1096 * On OSes not supporting symlinks we expect that RECENT.recent contains
1097 the contents of the principal recentfile. Actually this is identical on
1098 systems supporting symlinks. Simple, what follows from that is that we
1099 need to keep the serializer in the metadata because we cannot read it
1100 from the filename, doesn't it? Of course not. It's a chicken and egg
1101 problem. This leaves us with the problem to actually parse the
1102 serialized data to find out in which format it is. So who can do the 4
1103 or 5 magics we wanted to support? File::LibMagic?
1105 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
1107 * remotebase and recentfile_basename are ugly names. Now that we need a
1108 word for the shortest/principal/driving recentfile too we should do
1109 something about it.
1111 localroot is good. rfile is good. local_path() is bad, local_path($path)
1112 is medium, filenameroot() is bad, remotebase is bad, recentfile is
1113 already deprecated.
1115 Up to now remotebase was the string that described the remote root
1116 directory in rsync notation, like pause.perl.org::authors. And
1117 recentfile_basename was "RECENT-1h.yaml".
1119 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
1121 * The test that was added in today's checkin is a good start for a test
1122 of rmirror. We should have more methods in Recent.pm: verify,
1123 addmissingfiles. We should verify the current tree, then rmirror it and
1124 then verifytree the copy. We could then add some arbitrary file and let
1125 it be discovered by addmissingfiles, then rmirror again and then
1126 verifytree the copy again.
1128 Then we could start stealing from csync2 sqlite database [no port to
1129 OSX!] and fill a local DB. And methods to compare the database with the
1130 recentfiles. Our strength is that in principle we could maintain state
1131 with a single float. We have synced up to 1234567890.123456. If the Z
1132 file does not add new files all we have to do is mirror the new ones and
1133 delete the goners.
1135 This makes it clear that we should extend current protocol and declare
1136 that we cheat when we add files too late, just to help the other end
1137 keeping track. Ah yes, that's what was meant when zloop was mentioned
1138 earlier.
1140 Maybe need to revisit File::Mirror to help me with this task.
1142 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
1144 * There must be an allow-me-to-truncate flag in every recentfile.
1145 Without it one could construct a sequence of updates winning the locking
1146 battle against the aggregator. Only if an aggregator has managed to
1147 merge data over to the next level, truncating can be allowed. DONE with
1148 accessor merged.
1150 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
1152 * We should probably guarantee that no duplicates enter the aggregator
1153 array.
1155 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1157 * To get merge operation faster would need a good benchmark test. What
1158 02 spits out isn't reliable enough and is dominated by many other
1159 things. Between
1161 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
1162 Author: Andreas J. Koenig <k@k75.(none)>
1163 Date: Sat Aug 2 07:58:04 2008 +0200
1165 and
1167 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
1168 Author: Andreas J. Koenig <k@k75.(none)>
1169 Date: Sat Aug 2 11:40:29 2008 +0200
1171 there was a lot of trying but the effect is hardly measurable with
1172 current tests.
1174 * overhead of connecting seems high. When setting
1175 max_files_per_connection to 1 we see that.
1177 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1179 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
1181 25138.2215600014/3600 = 6.98283932222261
1183 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
1184 be a bug in the aggregator. FIXED
1186 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1188 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
1189 we have it, can we mirror it, mirror it!
1191 I fear this needs a new class which might be called
1192 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
1193 and treat them as an entity. I realize that a single recentfile may be
1194 sufficient for certain tasks and that it is handy for the low level
1195 programmer but it is not nice to use. If there is a delete in the 1h
1196 file then the 6h file still contains it. Seekers of the best information
1197 need to combine at least some of the recentfiles most of the time.
1199 There is the place for the Z loop!
1201 But the combination is something to collect in a database, isn't it. Did
1202 csync2 just harrumph?
1204 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1206 * it just occurred to me that hosts in the same mirroring pool could
1207 help out each other even without rewriting the recentfile. Just fetch
1208 the stuff to mirror from several places, bingo. But that's something
1209 that should rather live in a separate package or in rsync directly.
1211 * cronjobs are unsuited because with ntp they would all come at the full
1212 minute and disturb each other. Besides that I'd hate to have a backbone
1213 with more than a few seconds latency.
1215 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1217 * a second rsync server with access control for PAUSE. Port? 873 is the
1218 standard port, let's take 8873.
1220 * if there were a filesystem based on this, it would have a slow access
1221 to inexistent files. It would probably provide wrong readdir (only based
1222 on current content) or also a slow one (based on a recentfile written
1223 after the call). But it would provide fast access to existing files. Or
1224 one would deliberately allow slightly blurred answers based on some
1225 sqlite reflection of the recentfiles.
1227 * todo: write a variant of mirror() that combines two or more
1228 recentfiles and treats them like one
1230 * todo: signal handler to remove the tempfile
1232 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1234 * now that we have the symlink I forgot how it should be used in
1235 practice.
1237 * the z loop: add missing files to Z file. Just append them (instead of
1238 prepending). So one guy prepends something from the Y file from time to
1239 time and another guy appends something rather frequently. Collecting
1240 pond. When Y merges into Z, things get epoch and the collecting pond
1241 gets smaller. What exactly are "missing files"?
1243 take note of current epoch of the alpha file, let's call it the
1244 recent-ts
1246 find all files on disk
1248 remove all files registered in the recentworld up to recent-ts
1250 remove all files that have been deleted after recent-ts according to
1251 recentworld
1253 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1255 * rersyncrecent might be a cronjob with a (locked) state file which
1256 contains things like after and maybe last z sync or such?
1258 rrr-mirror might be an alternative name but how would we justify the
1259 three Rs when there is no Re-Rsync-Recent?
1261 With the --loop parameter it is an endless loop, without it is no loop.
1262 At least this is simple.
1264 * todo: new accssor z-interval specifies how often the Z file is updated
1265 against the filesystem. We probably want no epoch stamp on these
1266 entries. And we want to be able to filter the entries (e.g. no
1267 by-modules and by-category tree)
1269 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1271 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
1272 into Z. We do not need the other files filled up with filesystem
1273 contents though.
1275 * need interface to query for a file in order to NOT call update on
1276 PAUSE a second time within a short time.
1278 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
1280 * recommended update interval? Makes no sense, is different for
1281 different users.
1283 * Moosify
1285 Local Variables:
1286 mode: change-log
1287 change-log-default-name: "Todo"
1288 tab-width: 2
1289 left-margin: 2
1290 End: