1 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * I think, Done::register_one is doing wrong in that it does not
4 conflate neighboring pieces. The covered() method cannot do this because
5 it has no recent_events array at hand. But register_one has it and could
6 do it and for some reason misses to do it (sometimes).
8 This means that the three tests I just wrote can probably not survive
9 because they test with an already broken Done structure.
11 The art now is to detect how it happens, then to reproduce, then write a
14 So from the logfile this is what happens: we have a good interval with
15 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
16 top of it. Locally we now mirror F2 and open a new done interval for it.
17 Then we mirror F1 but this time with the timestamp T1b. And when we then
18 try to close the gap, we do not find T1 but instead something older. We
19 should gladly accept this older piece and this would fix this bug.
23 * bug to fix: when the 1h file changes while rmirror is running, we do
24 correctly sync the new files but never switch to the 6h file but rather
25 stay in a rather quick loop that fetches the 1h file again and again.
27 Is it possible that we initialize a new object? Or does
28 get_remote_recentfile_as_tempfile overwrite something in myself?
30 Want a new option: _runstatusfile => $file which frequently dumps the
31 state of all recentfiles to a file.
33 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
35 * Todo: now teach update to verify the timestamp is about to write
36 against the previous and use _increase_a_bit if it doesn't comply with
39 * The problem of rounding. So far perl's default precision was
40 sufficient. One day it won't be. FakeFloat has an easy job when it is
41 only reading and other machines have written correctly. But when we want
42 to write a floating point number that is a bit larger than the other
43 one, then we need our own idea of precision.
45 Slaven said: just append a "1". This might be going towards the end of
46 usability too quickly. I'd like something that actually uses the decimal
47 system. Well, appending a 1 also does this but...
49 E.g. we have 1.0. nextup on this architecture is starting with
50 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
51 taken the 1.0000000000000003 and the next user comes and the time tells
52 him 1.0 again. He has to beat my number without stepping over the
53 nextup. This is much less space than I had when I chose 1,2,3.
55 What is also irritating is that nextup is architecture dependent. The
56 128 bit guy must choose very long numbers to fit in between whereas the
57 other one with 16 bit uses larger steps. But then the algorithm is the
58 same for both, so that would be a nice thing.
60 I see two situation where we need this. One is when Time::HiRes returns
61 us a value that is <= the last entry in our recentfile. In this case
62 (let's call it the end-case) we must fill the region between that number
63 and the next higher native floating point number. The other is when we
64 inject an old file into an old recentfile (we would then also set a new
65 dirtymark). We find the integer value already taken and need a slightly
66 different one (let's call it the middle-case). The difference between
67 the two situations is that the next user will want to find something
68 higher than my number in the end-case and something lower than my number
71 So I suggest we give the function both a value and an upper bound and it
72 calculates us a primitive middle. The upper bound in the middle-case is
73 the next integer. The upper bound on the end-case is the nextup floating
74 point number. But the latter poses another problem: if we have occupied
75 the middle m between x and nextup(x), then the nextup(m) will probably
76 not be the same as nextup(x) because some rounding will take place
77 before the nextup is calculated and when the rounding reaches the
78 nextup(x), we will end up at nextup(nextup(x)).
80 So we really need to consider the nextup and the nextdown from there and
81 then the middle and that's the number we may approach asymptotically.
84 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
86 * consider deprecating the use of RECENT.recent as a symlink. It turns
87 out to need extra hoops with the rsync options and just isn't worth it.
88 Or maybe these extra hoops are needed anyway for the rest of the tree?
89 Nope, can't be the case because not all filesystems support symlinks.
91 But before doing the large step, I'll deprecate the call of
92 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
93 only introduced to resolve RECENT.recent and complicates the routine far
94 beyond what it deserves.
96 DONE. Won't deprecate RECENT.recent, just moved its handling to the
99 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
101 * I think it's a bug that the rsync_option links must be set to true in
102 order to support RECENT.recent and that nobody cares to set it
103 automatically. Similar for ignore_link_stat_errors.
105 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
107 * Todo: find all todos together and make a plan what is missing for a
110 - verifytree or something like that. fsck maybe.
112 - rersyncrecent, the script itself
114 - a way to only mirror the recentfiles without mirroring the whole
115 remote system such that people can decide to mirror only partially see
116 also 2008-08-30. .shadow-xxx directory? this also needed for a
117 filesystem that is still incomplete and might need the mirrorfiles for
120 - long living objects that mirror again and again. Inject something
121 into ta, see how it goes over to tb.
123 - how do we continue filling up the DONE system when we use an object
124 for the second time? "fully covered" and "uptodate" or new terminology.
126 - idea: have a new flag on recentfiles with the meaning: if this
127 changes, you're required to run a full rsync over all the files. The
128 reason why we set it would probably be: some foul happened. we injected
129 files in arbitrary places or didn't inject them although they changed.
130 The content of the flag? Timestamp? The relation between the
131 recentfiles would have to be inheritance from the principal, because any
132 out of band changes would soon later propagate to the next recentfile.
134 By upping the flag often one can easily ruin the slaves.
136 last out of band change? dirtymark?
138 Anyway, this implies that we read a potentially existing recentfile
141 And it implies that we have an eventloop that keeps us busy in 2-3
142 cycles, one for current stuff (tight loop) and one for the recentfiles
143 (cascade when principal has changed), one for the old stuff after a
146 - overview called on the wrong file should be understandable
148 - See that the long running process really only updates the principal
149 file unless it has missed a timespan during which something happened. If
150 nothing happened, it must notice even when it misses the timespan.
152 - fill up recentfiles with fake (historic) entries; fill up with
153 individual corrections; algorithm maybe to be done with bigfloat so that
154 we can always place something in the middle between two entries. Before
155 we must switch to bigfloat we could try to use Data::Float::nextup to
158 - the meta data field that must change when we fake something up so that
159 the downstream people know they have to re-fetch everything.
161 - how tolerant are we against missing files upstream? how do we keep
162 track? there are legitimate cases where we did read upstream index right
163 before a file got deleted there and then find that file as new and want
164 it. There are other cases that are not self healing and must be tracked
167 - how, exactly, do we have to deal with deletes? With rsync errors?
169 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
170 authors) failed: No such file or directory (2)
172 The file above is a delete in 1h and a new in file 1M and the
173 delete in the locally running rmirror did not get propagated to the 1M
174 object. Bug. And the consequence is a standstill.
176 It seems that a slave that works with a file below the principal needs
177 to merge things all the way up to get rid of later deletes. Or keep
178 track of all deletes and skip them later. So we need a trackdeletes.pm
179 similar to the done.pm?
181 see also 2008-08-20 about spurious deletes that really have no add
182 counterpart and yet they are not wrong.
184 - consider the effect when resyncing the recentfile takes longer than
185 the time per loop. Then we never rsync any file. We need to diagnose
186 that and force an increase of that loop time. But when we later are fast
187 enough again because the net has recovered, then we need to switch back
188 to original parameters. ERm, no, it's enough to keep syncing at least
189 one file before refetching an index file.
191 - remember to verify that no temp files are left lying around and the
194 - we must throw away the pathdb when we have reached the end of Z. From
195 that moment we can have a very small pathdb because the only reason for
196 a pathdb is that we know to ignore old records in old files. We won't
197 need this pathdb again before the next full pass over the data is
198 necessary and then we will rebuild it as we go along.
200 - status file for not long running jobs that want to track upstream with
203 - revisit all XXX _float areas and study Sub::Exporter DONE
205 - persistent DB even though we just said we do not need it. Just for
206 extended capabilities and time savings when, for example, upstream
207 announces a reset and we get new recentfiles and could then limit
208 ourselves to a subset of files (those that have a changed epoch) in a
209 first pass and would only then do the loop to verify the rest. Or
212 - lookup by epoch and by path and use this ability on the pause to never
213 again register a file twice that doesn't need it.
215 * Todo: aggregate files should know their feed and finding the principal
216 should be done stepwise. (?)
218 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
219 left locks around. DONE
221 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
223 * maybe extend the _overview so that it always says if and where the
224 last file is in the next file and where the next event in the next rf
225 would lie. No, don't like this anymore. REJECT
227 * take the two new redundant tests out again, only the third must
230 * Todo: add a sanity check if the merged structure is really pointing to
231 a different rf and that this different rf is larger. DONE
233 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
235 * now test, if they are overlapping. And test if there is a file in the
236 next rf that would fit into this rf's interval.
238 1h 1222324012.8474 1222322541.7963 0.4086
239 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
240 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
241 1W 1222313218.3626 1221708477.5829 167.9835
243 I suspect that somebody writes a merged timestamp without having merged
244 and then somebody else relies on it.
246 If aggregate is running, the intervals must not be extravagated, if it
247 is not running, there must not be bounds, the total number of events in
248 the system must be counted and must be controlled throughout the tests.
249 That the test required the additional update was probably nonsense,
250 because aggregate can cut pieces too. FIXED & DONE
252 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
254 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
257 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
259 * Most apparent bug at the moment is that the recentfiles are fetched
260 too often. Only the principal should be fetched and if it has not
261 changed, the others should not be refetched. ATM I must admit that I'm
262 happy that we refetch more often than needed because I can more easily
263 fix bugs while the thing is running.
265 * Let's say, 1220474966.19501 is a timestamp of a file that is already
266 done but the done system does not know about it. The reason for the
267 failure is not known and we never reach the status uptodate because of
268 this. We must get over it.
270 Later it turns out that the origin server had a bug somewhere.
271 1220474966.19042 came after 1220474966.19501. Or better: it was in the
272 array of the recentfile one position above. The bug was my own.
274 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
276 * There is the race condition where the server does a delete and the
277 slave does not yet know and then tries to download it because he sees
278 the new. So for this time window we must be more tolerant against
279 failure. If we cannot download a file, we should just skip it and should
280 not retry immediately. The whole system should discover the lost thing
281 later. Keeping track with the DONE system should really be a no brainer.
283 But there is something more: the whole filesystem is a database and the
284 recentfiles are one possible representation of it. It's a pretty useful
285 representation I think that's why I have implemented something around
286 it. But for strictly local operation it has little value. For local
287 operation we would much rather have a database. So we would enter every
288 recentfile reading and every rsync operation and for every file the last
289 state change and what it leads to. Then we would always ignore older
290 records without the efforts involved with recentfiles.
292 The database would have: path,recentepoch,rsyncedon,deletedon
294 Oh well, not yet clear where this leads to.
296 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
298 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
299 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
301 * statusfile, maybe only on demand, alone to have a sharp debugging
302 tool. It is locked and all recentfiles dump themselves into it and we
303 can build a viewer that lets us know where we stand and what's inside.
305 * remember: only the principal recentfile needs expiration, all others
306 shall be expired by principal if it discovers that something has move
309 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
311 * Always check if we stringify to a higher value than in the entry
314 * And in covered make an additional check if we would be able to see a
315 numerical difference between the two numbers and if we can't then switch
316 to a different, more expensive algorithm. Do not want to be caught by
317 floating surprises. DONE
319 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
321 * caching has several aspects here: we can cache the interval of the
322 recentfile which only will change when the mtime of the file changes. We
323 must re-mirror the recentfile when its ttl has expired. Does have_read
324 tell you anything? It counts nothing at all. Only the mtime is
325 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
326 thing the have_mirrored because it is unrelated to the mtime.
328 * Robustness of floating point calculations! I always thought that the
329 string calculated by the origin server for the floating representation
330 of the epoch time is just a string. When we convert it to a number and
331 later back to a string, the other computer might come to a different
332 conclusion. This must not happen, we want to preserve it under any
333 circumstances. I will have to write tests with overlong sequences that
334 get lost in arithmetic and must see if all still works well. DONE
336 But one fragile point remains: if one host considers a>b and the other
337 one considers them == but no eq. To prevent this, we must probably do
338 some extra homework. DONE
340 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
342 * the concept of tracking DONE needs an object per recentfile that has
343 something like these methods:
345 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
347 covered() register() covered()
349 The unclear thing is how we translate points in time into intervals. We
350 could pass a reference to the current recent_events array when running
351 we_have(xxx) and let the DONE object iterate over it such that it only
352 has to store a list of intervals that can melt into each other. Ah, even
353 passing the list together with a list of indexes seems feasiable.
355 Or maybe ask for the inverted list?
357 Whenever the complete array is covered by the interval we say we are
358 fully covered and if the recentfile is not expired, we are uptodate.
360 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
362 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
364 * need a way to "return" the next entry after the end of a list. When
365 the caller says "before" or "after" we would like to know if he could
366 cover that interval/threshold or not because this influences the effect
367 of a newer timestamp of that recentfile. DONE with $opt{info}.
369 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
371 * one of the next things to tackle: the equivalent of csync2 -TIXU.
373 loop implies tixu (?). Nope, something like --statefile decides. Per
376 T test, I init, X including removals, U nodirtymark
378 So we have no concept of dirtymarks, we only trust that since we are
379 running we have observed everything steadily. But people will not let
380 this program run forever so we must consider both startup penalty and
381 book keeping for later runs. We keep this for later. For now we write a
382 long running mirror that merges several intervals.
384 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
386 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
389 * rersyncrecent, the script: default to one week. The name of the switch
390 is --after. Other switches? --loop!
392 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
394 * need a switch --skip-deletes (?)
396 * need a switch --enduser that tells us that the whole tempfile
397 discipline is not needed when there is no downstream user. (?)
399 Without this switch we cannot have a reasonable recent.pl that just
400 displays the recent additions. Either we accept to download everything.
401 Or we download temporary files without the typical rsync protocol
404 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
405 File::Temp::tempdir, this might be a win.
407 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
409 * apropos missing: we have no push, we never know the downstream
410 servers. People who know their downstream hosts and want to ascertain
411 something will want additional methods we have never thought about, like
412 update or delete a certain file.
414 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
416 * tempted to refactor rmirror into resolve_symlink, localize, etc.
417 Curious if rsync_options=links equal 0 vs. 1 will make the expected
420 * rsync options: it's a bit of a pain that we usually need several rsync
421 options, like compress, links, times, checksum and that there is no
422 reasonable default except the original rsync default. I think wee can
423 safely assume that the rsync options are shared between all recentfile
424 instances within one recent tree.
426 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
428 * deletes: if a delete follows an add quickly enough it may happen that
429 a downstream mirror did not see the add at all! It seems this needs to
430 be mentioned somewhere. The point here is that even if the downstream is
431 never missing the principal timeframe it may encounter a "delete" that
432 has no complimentary "add" anywhere.
434 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
436 * I suspect the treat of metadata is incorrect during read or something.
437 The bug that I am watching is that between 06:08 and 06:09 the 6h file
438 contained more than 6 hours worth of data. At 06:08 we merged into the
439 1d file. We need to take snapshots of the 6h file over the course of an
440 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
443 Much worse: watching the 1h file: right at the moment (at 06:35) it
444 covers 1218867584-1219120397 which is 70 hours.
446 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
447 08:19:44 2008, that is when I checked out last time, so it seems to be
448 aggregating and never truncating?
450 No, correct is: it is never truncating; but wrong is: it is aggregating.
451 It does receive a lot of events from time to time from a larger file.
452 Somehow a large file gets merged into the small one and because the
453 "meta/merged" attribute is missing, nobody is paying attention. I
454 believe that I can fix this by making sure that metadata are honoured
455 during read. DONE and test adjusted.
457 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
459 * grand renaming plan
461 remotebase => remoteroot to fit well with localroot DONE
462 local_path() => localroot seems to me should already work DONE
463 recentfile_basename => rfilename no need to stress it has no slash DONE
465 filenameroot??? Doesn't seem too bad to me today. Maybe something like
466 kern? It would anyway need a deprecation cycle because it is an
467 important constructor.
469 * I like the portability that Data::Serializer brings us but the price
470 is that some day we might find out that it is slowing us a bit. We'll
473 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
475 * should we not enter the interval of the principal (or the interval of
476 the merging file?) in every aggregated/merged file?
478 * we should aim at a first release and give up on thinking about
479 sanitizing stuff and zloop. Let's just admit that a full traditional
480 rsync is the only available sanitizer ATM. Otherwise it's complicated
481 stuff: sanitizing on the origin server, sanitizing on the slaves,
482 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
483 get the basics out before this becomes a major cause for mess.
485 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
487 * On OSes not supporting symlinks we expect that RECENT.recent contains
488 the contents of the principal recentfile. Actually this is identical on
489 systems supporting symlinks. Simple, what follows from that is that we
490 need to keep the serializer in the metadata because we cannot read it
491 from the filename, doesn't it? Of course not. It's a chicken and egg
492 problem. This leaves us with the problem to actually parse the
493 serialized data to find out in which format it is. So who can do the 4
494 or 5 magics we wanted to support? File::LibMagic?
496 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
498 * remotebase and recentfile_basename are ugly names. Now that we need a
499 word for the shortest/principal/driving recentfile too we should do
502 localroot is good. rfile is good. local_path() is bad, local_path($path)
503 is medium, filenameroot() is bad, remotebase is bad, recentfile is
506 Up to now remotebase was the string that described the remote root
507 directory in rsync notation, like pause.perl.org::authors. And
508 recentfile_basename was "RECENT-1h.yaml".
510 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
512 * The test that was added in today's checkin is a good start for a test
513 of rmirror. We should have more methods in Recent.pm: verify,
514 addmissingfiles. We should verify the current tree, then rmirror it and
515 then verifytree the copy. We could then add some arbitrary file and let
516 it be discovered by addmissingfiles, then rmirror again and then
517 verifytree the copy again.
519 Then we could start stealing from csync2 sqlite database [no port to
520 OSX!] and fill a local DB. And methods to compare the database with the
521 recentfiles. Our strength is that in principle we could maintain state
522 with a single float. We have synced up to 1234567890.123456. If the Z
523 file does not add new files all we have to do is mirror the new ones and
526 This makes it clear that we should extend current protocol and declare
527 that we cheat when we add files too late, just to help the other end
528 keeping track. Ah yes, that's what was meant when zloop was mentioned
531 Maybe need to revisit File::Mirror to help me with this task.
533 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
535 * There must be an allow-me-to-truncate flag in every recentfile.
536 Without it one could construct a sequence of updates winning the locking
537 battle against the aggregator. Only if an aggregator has managed to
538 merge data over to the next level, truncating can be allowed. DONE with
541 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
543 * We should probably guarantee that no duplicates enter the aggregator
546 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
548 * To get merge operation faster would need a good benchmark test. What
549 02 spits out isn't reliable enough and is dominated by many other
552 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
553 Author: Andreas J. Koenig <k@k75.(none)>
554 Date: Sat Aug 2 07:58:04 2008 +0200
558 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
559 Author: Andreas J. Koenig <k@k75.(none)>
560 Date: Sat Aug 2 11:40:29 2008 +0200
562 there was a lot of trying but the effect is hardly measurable with
565 * overhead of connecting seems high. When setting
566 max_files_per_connection to 1 we see that.
568 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
570 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
572 25138.2215600014/3600 = 6.98283932222261
574 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
575 be a bug in the aggregator. FIXED
577 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
579 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
580 we have it, can we mirror it, mirror it!
582 I fear this needs a new class which might be called
583 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
584 and treat them as an entity. I realize that a single recentfile may be
585 sufficient for certain tasks and that it is handy for the low level
586 programmer but it is not nice to use. If there is a delete in the 1h
587 file then the 6h file still contains it. Seekers of the best information
588 need to combine at least some of the recentfiles most of the time.
590 There is the place for the Z loop!
592 But the combination is something to collect in a database, isn't it. Did
593 csync2 just harrumph?
595 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
597 * it just occurred to me that hosts in the same mirroring pool could
598 help out each other even without rewriting the recentfile. Just fetch
599 the stuff to mirror from several places, bingo. But that's something
600 that should rather live in a separate package or in rsync directly.
602 * cronjobs are unsuited because with ntp they would all come at the full
603 minute and disturb each other. Besides that I'd hate to have a backbone
604 with more than a few seconds latency.
606 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
608 * a second rsync server with access control for PAUSE. Port? 873 is the
609 standard port, let's take 8873.
611 * if there were a filesystem based on this, it would have a slow access
612 to inexistent files. It would probably provide wrong readdir (only based
613 on current content) or also a slow one (based on a recentfile written
614 after the call). But it would provide fast access to existing files. Or
615 one would deliberately allow slightly blurred answers based on some
616 sqlite reflection of the recentfiles.
618 * todo: write a variant of mirror() that combines two or more
619 recentfiles and treats them like one
621 * todo: signal handler to remove the tempfile
623 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
625 * now that we have the symlink I forgot how it should be used in
628 * the z loop: add missing files to Z file. Just append them (instead of
629 prepending). So one guy prepends something from the Y file from time to
630 time and another guy appends something rather frequently. Collecting
631 pond. When Y merges into Z, things get epoch and the collecting pond
632 gets smaller. What exactly are "missing files"?
634 take note of current epoch of the alpha file, let's call it the
637 find all files on disk
639 remove all files registered in the recentworld up to recent-ts
641 remove all files that have been deleted after recent-ts according to
644 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
646 * rersyncrecent might be a cronjob with a (locked) state file which
647 contains things like after and maybe last z sync or such?
649 rrr-mirror might be an alternative name but how would we justify the
650 three Rs when there is no Re-Rsync-Recent?
652 With the --loop parameter it is an endless loop, without it is no loop.
653 At least this is simple.
655 * todo: new accssor z-interval specifies how often the Z file is updated
656 against the filesystem. We probably want no epoch stamp on these
657 entries. And we want to be able to filter the entries (e.g. no
658 by-modules and by-category tree)
660 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
662 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
663 into Z. We do not need the other files filled up with filesystem
666 * need interface to query for a file in order to NOT call update on
667 PAUSE a second time within a short time.
669 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
671 * recommended update interval? Makes no sense, is different for
678 change-log-default-name: "Todo"