1 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * Slaven suggests to have the current epoch or the whole current
4 recentfile available from the HTTP server and take it away with
5 keepalive. This direction goes the granularity down to subseconds.
7 We might want to rewrite everything to factor out transport and allow
8 the whole thing to run via HTTP.
10 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
12 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
15 * smoker on k81 fetching from k75 to verify cascading works. See
16 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
18 * maybe the loop should wait for CHECKSUMS file after every upload. And
19 CPAN.pm needs to deal with timestamps in the future.
21 * do not forget the dirtymark!
23 Text: have a new flag on recentfiles with the meaning: if this
24 changes, you're required to run a full rsync over all the files. The
25 reason why we set it would probably be: some foul happened. we injected
26 files in arbitrary places or didn't inject them although they changed.
27 The content of the flag? Timestamp? The relation between the
28 recentfiles would have to be inheritance from the principal, because any
29 out of band changes would soon later propagate to the next recentfile.
31 By upping the flag often one can easily ruin the slaves.
33 last out of band change? dirtymark?
35 Anyway, this implies that we read a potentially existing recentfile
38 And it implies that we have an eventloop that keeps us busy in 2-3
39 cycles, one for current stuff (tight loop) and one for the recentfiles
40 (cascade when principal has changed), one for the old stuff after a
43 And it implies that the out-of-band change in any of the recentfiles
44 must have a lock on the principal file and there is the place to set the
47 * after the dirtymark is done: fill up recentfiles with fake (historic)
48 entries; fill up with individual corrections; algorithm maybe to be done
49 with bigfloat so that we can always place something in the middle
50 between two entries. Before we must switch to bigfloat we could try to
51 use Data::Float::nextup to get the.
53 * lookup by epoch and by path and use this ability on the pause to never
54 again register a file twice that doesn't need it.
56 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
59 * start a FAQ, especially quick start guide questions. Also to aid those
60 problematic areas where we have no good solution, like the "links"
63 * wish feedback when we are slow.
67 * Remove a few DEBUG statements.
69 * The multiple-rrr way of doing things needs a new option to rmirror,
70 like piecemeal or so. Not urgent because after the first pass through,
71 things run smoothely. It's only ugly during the first pass.
73 * I have the suspicion that the code is broken that decides if the
74 neighboring RF needs to be seeded. I fear when too much time has gone
75 between two calls (in our case more than one hour), it would not seed
76 the neighbor. Of course this will never be noticed, so we need a good
79 * local/localroot confusion: I currently pass both options but one must
82 * accounts for early birds on PAUSE rsync daemon.
84 * hardcoded 20 seconds
86 * who mirrors the index? DOING now.
88 * which CPAN mirrors offer rsync?
90 * visit all XXX, visit all _float places
92 * rename the pathdb stuff, it's too confusing. No idea how.
94 * rrr-inotify, backpan, rrr-register
96 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
98 * current bugs: the pathdb seems to get no reset, the seeding of the
99 secondaryttl stuff seems not to have an effect. Have helped myself with
100 a rand(10), need to fix this back. So not checked in. Does the rand
103 The rand thing helps. The secondaryttl stuff was in the wrong line,
106 The pathdb stuff was because I called either _pathdb or __pathdb on the
107 wrong object. FIXED now.
109 * It's not so beautiful if we never fetch the recentfiles that are not
110 the principal, even if this is correct behaviour. We really do not need
111 them after we have fetched the whole content.
113 OK, we want a switch for that: secondaryttl DONE
115 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
117 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
118 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
119 it lets 40 follow. FIXED
121 * See that the long running process really only updates the principal
122 file unless it has missed a timespan during which something happened. If
123 nothing happened, it must notice even when it misses the timespan. DONE
125 * we must throw away the pathdb when we have reached the end of Z. From
126 that moment we can have a very small pathdb because the only reason for
127 a pathdb is that we know to ignore old records in old files. We won't
128 need this pathdb again before the next full pass over the data is
129 necessary and then we will rebuild it as we go along. DONE
131 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
133 * I think, Done::register_one is doing wrong in that it does not
134 conflate neighboring pieces. The covered() method cannot do this because
135 it has no recent_events array at hand. But register_one has it and could
136 do it and for some reason misses to do it (sometimes).
138 This means that the three tests I just wrote can probably not survive
139 because they test with an already broken Done structure.
141 The art now is to detect how it happens, then to reproduce, then write a
144 So from the logfile this is what happens: we have a good interval with
145 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
146 top of it. Locally we now mirror F2 and open a new done interval for it.
147 Then we mirror F1 but this time with the timestamp T1b. And when we then
148 try to close the gap, we do not find T1 but instead something older. We
149 should gladly accept this older piece and this would fix this bug.
153 * bug to fix: when the 1h file changes while rmirror is running, we do
154 correctly sync the new files but never switch to the 6h file but rather
155 stay in a rather quick loop that fetches the 1h file again and again.
157 Is it possible that we initialize a new object? Or does
158 get_remote_recentfile_as_tempfile overwrite something in myself?
160 Want a new option: _runstatusfile => $file which frequently dumps the
161 state of all recentfiles to a file.
165 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
167 * Todo: now teach update to verify the timestamp is about to write
168 against the previous and use _increase_a_bit if it doesn't comply with
169 strict monotony. DONE
171 * The problem of rounding. So far perl's default precision was
172 sufficient. One day it won't be. FakeFloat has an easy job when it is
173 only reading and other machines have written correctly. But when we want
174 to write a floating point number that is a bit larger than the other
175 one, then we need our own idea of precision.
177 Slaven said: just append a "1". This might be going towards the end of
178 usability too quickly. I'd like something that actually uses the decimal
179 system. Well, appending a 1 also does this but...
181 E.g. we have 1.0. nextup on this architecture is starting with
182 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
183 taken the 1.0000000000000003 and the next user comes and the time tells
184 him 1.0 again. He has to beat my number without stepping over the
185 nextup. This is much less space than I had when I chose 1,2,3.
187 What is also irritating is that nextup is architecture dependent. The
188 128 bit guy must choose very long numbers to fit in between whereas the
189 other one with 16 bit uses larger steps. But then the algorithm is the
190 same for both, so that would be a nice thing.
192 I see two situation where we need this. One is when Time::HiRes returns
193 us a value that is <= the last entry in our recentfile. In this case
194 (let's call it the end-case) we must fill the region between that number
195 and the next higher native floating point number. The other is when we
196 inject an old file into an old recentfile (we would then also set a new
197 dirtymark). We find the integer value already taken and need a slightly
198 different one (let's call it the middle-case). The difference between
199 the two situations is that the next user will want to find something
200 higher than my number in the end-case and something lower than my number
203 So I suggest we give the function both a value and an upper bound and it
204 calculates us a primitive middle. The upper bound in the middle-case is
205 the next integer. The upper bound on the end-case is the nextup floating
206 point number. But the latter poses another problem: if we have occupied
207 the middle m between x and nextup(x), then the nextup(m) will probably
208 not be the same as nextup(x) because some rounding will take place
209 before the nextup is calculated and when the rounding reaches the
210 nextup(x), we will end up at nextup(nextup(x)).
212 So we really need to consider the nextup and the nextdown from there and
213 then the middle and that's the number we may approach asymptotically.
216 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
218 * consider deprecating the use of RECENT.recent as a symlink. It turns
219 out to need extra hoops with the rsync options and just isn't worth it.
220 Or maybe these extra hoops are needed anyway for the rest of the tree?
221 Nope, can't be the case because not all filesystems support symlinks.
223 But before doing the large step, I'll deprecate the call of
224 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
225 only introduced to resolve RECENT.recent and complicates the routine far
226 beyond what it deserves.
228 DONE. Won't deprecate RECENT.recent, just moved its handling to the
231 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
233 * I think it's a bug that the rsync_option links must be set to true in
234 order to support RECENT.recent and that nobody cares to set it
235 automatically. Similar for ignore_link_stat_errors.
237 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
239 * Todo: find all todos together and make a plan what is missing for a
242 - verifytree or something like that. fsck maybe.
244 - rersyncrecent, the script itself? What it do?
246 - a way to only mirror the recentfiles without mirroring the whole
247 remote system such that people can decide to mirror only partially see
248 also 2008-08-30. .shadow-xxx directory? this also needed for a
249 filesystem that is still incomplete and might need the mirrorfiles for
252 - long living objects that mirror again and again. Inject something
253 into ta, see how it goes over to tb.
255 - how do we continue filling up the DONE system when we use an object
256 for the second time? "fully covered" and "uptodate" or new terminology.
258 - overview called on the wrong file should be understandable
260 - the meta data field that must change when we fake something up so that
261 the downstream people know they have to re-fetch everything.
263 - how tolerant are we against missing files upstream? how do we keep
264 track? there are legitimate cases where we did read upstream index right
265 before a file got deleted there and then find that file as new and want
266 it. There are other cases that are not self healing and must be tracked
269 - how, exactly, do we have to deal with deletes? With rsync errors?
271 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
272 authors) failed: No such file or directory (2)
274 The file above is a delete in 1h and a new in file 1M and the
275 delete in the locally running rmirror did not get propagated to the 1M
276 object. Bug. And the consequence is a standstill.
278 It seems that a slave that works with a file below the principal needs
279 to merge things all the way up to get rid of later deletes. Or keep
280 track of all deletes and skip them later. So we need a trackdeletes.pm
281 similar to the done.pm?
283 see also 2008-08-20 about spurious deletes that really have no add
284 counterpart and yet they are not wrong.
286 - consider the effect when resyncing the recentfile takes longer than
287 the time per loop. Then we never rsync any file. We need to diagnose
288 that and force an increase of that loop time. But when we later are fast
289 enough again because the net has recovered, then we need to switch back
290 to original parameters. ERm, no, it's enough to keep syncing at least
291 one file before refetching an index file.
293 - remember to verify that no temp files are left lying around and the
296 - status file for not long running jobs that want to track upstream with
299 - revisit all XXX _float areas and study Sub::Exporter DONE
301 - persistent DB even though we just said we do not need it. Just for
302 extended capabilities and time savings when, for example, upstream
303 announces a reset and we get new recentfiles and could then limit
304 ourselves to a subset of files (those that have a changed epoch) in a
305 first pass and would only then do the loop to verify the rest. Or
308 * Todo: aggregate files should know their feed and finding the principal
309 should be done stepwise. (?)
311 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
312 left locks around. DONE
314 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
316 * maybe extend the _overview so that it always says if and where the
317 last file is in the next file and where the next event in the next rf
318 would lie. No, don't like this anymore. REJECT
320 * take the two new redundant tests out again, only the third must
323 * Todo: add a sanity check if the merged structure is really pointing to
324 a different rf and that this different rf is larger. DONE
326 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
328 * now test, if they are overlapping. And test if there is a file in the
329 next rf that would fit into this rf's interval.
331 1h 1222324012.8474 1222322541.7963 0.4086
332 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
333 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
334 1W 1222313218.3626 1221708477.5829 167.9835
336 I suspect that somebody writes a merged timestamp without having merged
337 and then somebody else relies on it.
339 If aggregate is running, the intervals must not be extravagated, if it
340 is not running, there must not be bounds, the total number of events in
341 the system must be counted and must be controlled throughout the tests.
342 That the test required the additional update was probably nonsense,
343 because aggregate can cut pieces too. FIXED & DONE
345 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
347 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
350 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
352 * Most apparent bug at the moment is that the recentfiles are fetched
353 too often. Only the principal should be fetched and if it has not
354 changed, the others should not be refetched. ATM I must admit that I'm
355 happy that we refetch more often than needed because I can more easily
356 fix bugs while the thing is running.
358 * Let's say, 1220474966.19501 is a timestamp of a file that is already
359 done but the done system does not know about it. The reason for the
360 failure is not known and we never reach the status uptodate because of
361 this. We must get over it.
363 Later it turns out that the origin server had a bug somewhere.
364 1220474966.19042 came after 1220474966.19501. Or better: it was in the
365 array of the recentfile one position above. The bug was my own.
367 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
369 * There is the race condition where the server does a delete and the
370 slave does not yet know and then tries to download it because he sees
371 the new. So for this time window we must be more tolerant against
372 failure. If we cannot download a file, we should just skip it and should
373 not retry immediately. The whole system should discover the lost thing
374 later. Keeping track with the DONE system should really be a no brainer.
376 But there is something more: the whole filesystem is a database and the
377 recentfiles are one possible representation of it. It's a pretty useful
378 representation I think that's why I have implemented something around
379 it. But for strictly local operation it has little value. For local
380 operation we would much rather have a database. So we would enter every
381 recentfile reading and every rsync operation and for every file the last
382 state change and what it leads to. Then we would always ignore older
383 records without the efforts involved with recentfiles.
385 The database would have: path,recentepoch,rsyncedon,deletedon
387 Oh well, not yet clear where this leads to.
389 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
391 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
392 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
394 * statusfile, maybe only on demand, alone to have a sharp debugging
395 tool. It is locked and all recentfiles dump themselves into it and we
396 can build a viewer that lets us know where we stand and what's inside.
398 * remember: only the principal recentfile needs expiration, all others
399 shall be expired by principal if it discovers that something has move
402 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
404 * Always check if we stringify to a higher value than in the entry
407 * And in covered make an additional check if we would be able to see a
408 numerical difference between the two numbers and if we can't then switch
409 to a different, more expensive algorithm. Do not want to be caught by
410 floating surprises. DONE
412 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
414 * caching has several aspects here: we can cache the interval of the
415 recentfile which only will change when the mtime of the file changes. We
416 must re-mirror the recentfile when its ttl has expired. Does have_read
417 tell you anything? It counts nothing at all. Only the mtime is
418 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
419 thing the have_mirrored because it is unrelated to the mtime.
421 * Robustness of floating point calculations! I always thought that the
422 string calculated by the origin server for the floating representation
423 of the epoch time is just a string. When we convert it to a number and
424 later back to a string, the other computer might come to a different
425 conclusion. This must not happen, we want to preserve it under any
426 circumstances. I will have to write tests with overlong sequences that
427 get lost in arithmetic and must see if all still works well. DONE
429 But one fragile point remains: if one host considers a>b and the other
430 one considers them == but no eq. To prevent this, we must probably do
431 some extra homework. DONE
433 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
435 * the concept of tracking DONE needs an object per recentfile that has
436 something like these methods:
438 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
440 covered() register() covered()
442 The unclear thing is how we translate points in time into intervals. We
443 could pass a reference to the current recent_events array when running
444 we_have(xxx) and let the DONE object iterate over it such that it only
445 has to store a list of intervals that can melt into each other. Ah, even
446 passing the list together with a list of indexes seems feasiable.
448 Or maybe ask for the inverted list?
450 Whenever the complete array is covered by the interval we say we are
451 fully covered and if the recentfile is not expired, we are uptodate.
453 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
455 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
457 * need a way to "return" the next entry after the end of a list. When
458 the caller says "before" or "after" we would like to know if he could
459 cover that interval/threshold or not because this influences the effect
460 of a newer timestamp of that recentfile. DONE with $opt{info}.
462 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
464 * one of the next things to tackle: the equivalent of csync2 -TIXU.
466 loop implies tixu (?). Nope, something like --statefile decides. Per
469 T test, I init, X including removals, U nodirtymark
471 So we have no concept of dirtymarks, we only trust that since we are
472 running we have observed everything steadily. But people will not let
473 this program run forever so we must consider both startup penalty and
474 book keeping for later runs. We keep this for later. For now we write a
475 long running mirror that merges several intervals.
477 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
479 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
482 * rersyncrecent, the script: default to one week. The name of the switch
483 is --after. Other switches? --loop!
485 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
487 * need a switch --skip-deletes (?)
489 * need a switch --enduser that tells us that the whole tempfile
490 discipline is not needed when there is no downstream user. (?)
492 Without this switch we cannot have a reasonable recent.pl that just
493 displays the recent additions. Either we accept to download everything.
494 Or we download temporary files without the typical rsync protocol
497 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
498 File::Temp::tempdir, this might be a win.
500 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
502 * apropos missing: we have no push, we never know the downstream
503 servers. People who know their downstream hosts and want to ascertain
504 something will want additional methods we have never thought about, like
505 update or delete a certain file.
507 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
509 * tempted to refactor rmirror into resolve_symlink, localize, etc.
510 Curious if rsync_options=links equal 0 vs. 1 will make the expected
513 * rsync options: it's a bit of a pain that we usually need several rsync
514 options, like compress, links, times, checksum and that there is no
515 reasonable default except the original rsync default. I think wee can
516 safely assume that the rsync options are shared between all recentfile
517 instances within one recent tree.
519 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
521 * deletes: if a delete follows an add quickly enough it may happen that
522 a downstream mirror did not see the add at all! It seems this needs to
523 be mentioned somewhere. The point here is that even if the downstream is
524 never missing the principal timeframe it may encounter a "delete" that
525 has no complimentary "add" anywhere.
527 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
529 * I suspect the treat of metadata is incorrect during read or something.
530 The bug that I am watching is that between 06:08 and 06:09 the 6h file
531 contained more than 6 hours worth of data. At 06:08 we merged into the
532 1d file. We need to take snapshots of the 6h file over the course of an
533 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
536 Much worse: watching the 1h file: right at the moment (at 06:35) it
537 covers 1218867584-1219120397 which is 70 hours.
539 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
540 08:19:44 2008, that is when I checked out last time, so it seems to be
541 aggregating and never truncating?
543 No, correct is: it is never truncating; but wrong is: it is aggregating.
544 It does receive a lot of events from time to time from a larger file.
545 Somehow a large file gets merged into the small one and because the
546 "meta/merged" attribute is missing, nobody is paying attention. I
547 believe that I can fix this by making sure that metadata are honoured
548 during read. DONE and test adjusted.
550 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
552 * grand renaming plan
554 remotebase => remoteroot to fit well with localroot DONE
555 local_path() => localroot seems to me should already work DONE
556 recentfile_basename => rfilename no need to stress it has no slash DONE
558 filenameroot??? Doesn't seem too bad to me today. Maybe something like
559 kern? It would anyway need a deprecation cycle because it is an
560 important constructor.
562 * I like the portability that Data::Serializer brings us but the price
563 is that some day we might find out that it is slowing us a bit. We'll
566 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
568 * should we not enter the interval of the principal (or the interval of
569 the merging file?) in every aggregated/merged file?
571 * we should aim at a first release and give up on thinking about
572 sanitizing stuff and zloop. Let's just admit that a full traditional
573 rsync is the only available sanitizer ATM. Otherwise it's complicated
574 stuff: sanitizing on the origin server, sanitizing on the slaves,
575 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
576 get the basics out before this becomes a major cause for mess.
578 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
580 * On OSes not supporting symlinks we expect that RECENT.recent contains
581 the contents of the principal recentfile. Actually this is identical on
582 systems supporting symlinks. Simple, what follows from that is that we
583 need to keep the serializer in the metadata because we cannot read it
584 from the filename, doesn't it? Of course not. It's a chicken and egg
585 problem. This leaves us with the problem to actually parse the
586 serialized data to find out in which format it is. So who can do the 4
587 or 5 magics we wanted to support? File::LibMagic?
589 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
591 * remotebase and recentfile_basename are ugly names. Now that we need a
592 word for the shortest/principal/driving recentfile too we should do
595 localroot is good. rfile is good. local_path() is bad, local_path($path)
596 is medium, filenameroot() is bad, remotebase is bad, recentfile is
599 Up to now remotebase was the string that described the remote root
600 directory in rsync notation, like pause.perl.org::authors. And
601 recentfile_basename was "RECENT-1h.yaml".
603 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
605 * The test that was added in today's checkin is a good start for a test
606 of rmirror. We should have more methods in Recent.pm: verify,
607 addmissingfiles. We should verify the current tree, then rmirror it and
608 then verifytree the copy. We could then add some arbitrary file and let
609 it be discovered by addmissingfiles, then rmirror again and then
610 verifytree the copy again.
612 Then we could start stealing from csync2 sqlite database [no port to
613 OSX!] and fill a local DB. And methods to compare the database with the
614 recentfiles. Our strength is that in principle we could maintain state
615 with a single float. We have synced up to 1234567890.123456. If the Z
616 file does not add new files all we have to do is mirror the new ones and
619 This makes it clear that we should extend current protocol and declare
620 that we cheat when we add files too late, just to help the other end
621 keeping track. Ah yes, that's what was meant when zloop was mentioned
624 Maybe need to revisit File::Mirror to help me with this task.
626 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
628 * There must be an allow-me-to-truncate flag in every recentfile.
629 Without it one could construct a sequence of updates winning the locking
630 battle against the aggregator. Only if an aggregator has managed to
631 merge data over to the next level, truncating can be allowed. DONE with
634 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
636 * We should probably guarantee that no duplicates enter the aggregator
639 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
641 * To get merge operation faster would need a good benchmark test. What
642 02 spits out isn't reliable enough and is dominated by many other
645 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
646 Author: Andreas J. Koenig <k@k75.(none)>
647 Date: Sat Aug 2 07:58:04 2008 +0200
651 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
652 Author: Andreas J. Koenig <k@k75.(none)>
653 Date: Sat Aug 2 11:40:29 2008 +0200
655 there was a lot of trying but the effect is hardly measurable with
658 * overhead of connecting seems high. When setting
659 max_files_per_connection to 1 we see that.
661 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
663 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
665 25138.2215600014/3600 = 6.98283932222261
667 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
668 be a bug in the aggregator. FIXED
670 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
672 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
673 we have it, can we mirror it, mirror it!
675 I fear this needs a new class which might be called
676 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
677 and treat them as an entity. I realize that a single recentfile may be
678 sufficient for certain tasks and that it is handy for the low level
679 programmer but it is not nice to use. If there is a delete in the 1h
680 file then the 6h file still contains it. Seekers of the best information
681 need to combine at least some of the recentfiles most of the time.
683 There is the place for the Z loop!
685 But the combination is something to collect in a database, isn't it. Did
686 csync2 just harrumph?
688 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
690 * it just occurred to me that hosts in the same mirroring pool could
691 help out each other even without rewriting the recentfile. Just fetch
692 the stuff to mirror from several places, bingo. But that's something
693 that should rather live in a separate package or in rsync directly.
695 * cronjobs are unsuited because with ntp they would all come at the full
696 minute and disturb each other. Besides that I'd hate to have a backbone
697 with more than a few seconds latency.
699 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
701 * a second rsync server with access control for PAUSE. Port? 873 is the
702 standard port, let's take 8873.
704 * if there were a filesystem based on this, it would have a slow access
705 to inexistent files. It would probably provide wrong readdir (only based
706 on current content) or also a slow one (based on a recentfile written
707 after the call). But it would provide fast access to existing files. Or
708 one would deliberately allow slightly blurred answers based on some
709 sqlite reflection of the recentfiles.
711 * todo: write a variant of mirror() that combines two or more
712 recentfiles and treats them like one
714 * todo: signal handler to remove the tempfile
716 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
718 * now that we have the symlink I forgot how it should be used in
721 * the z loop: add missing files to Z file. Just append them (instead of
722 prepending). So one guy prepends something from the Y file from time to
723 time and another guy appends something rather frequently. Collecting
724 pond. When Y merges into Z, things get epoch and the collecting pond
725 gets smaller. What exactly are "missing files"?
727 take note of current epoch of the alpha file, let's call it the
730 find all files on disk
732 remove all files registered in the recentworld up to recent-ts
734 remove all files that have been deleted after recent-ts according to
737 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
739 * rersyncrecent might be a cronjob with a (locked) state file which
740 contains things like after and maybe last z sync or such?
742 rrr-mirror might be an alternative name but how would we justify the
743 three Rs when there is no Re-Rsync-Recent?
745 With the --loop parameter it is an endless loop, without it is no loop.
746 At least this is simple.
748 * todo: new accssor z-interval specifies how often the Z file is updated
749 against the filesystem. We probably want no epoch stamp on these
750 entries. And we want to be able to filter the entries (e.g. no
751 by-modules and by-category tree)
753 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
755 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
756 into Z. We do not need the other files filled up with filesystem
759 * need interface to query for a file in order to NOT call update on
760 PAUSE a second time within a short time.
762 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
764 * recommended update interval? Makes no sense, is different for
771 change-log-default-name: "Todo"