1 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * The multiple-rrr way of doing things needs a new option to rmirror,
4 like piecemeal or so. Not urgent because after the first pass through,
5 things run smoothely. It's only ugly during the first pass.
7 * I have the suspicion that the code is broken that decides if the
8 neighboring RF needs to be seeded. I fear when too much time has gone
9 between two calls (in our case more than one hour), it would not seed
10 the neighbor. Of course this will never be noticed, so we need a good
13 * local/localroot confusion
15 * accounts for early birds
17 * hardcoded 20 seconds
19 * who mirrors the index?
21 * which CPAN mirrors offer rsync?
23 * visit all XXX, visit all _float places
25 * rename the pathdb stuff, it's too confusing. No idea how.
27 * rrr-inotify, backpan, rrr-register
29 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
31 * current bugs: the pathdb seems to get no reset, the seeding of the
32 secondaryttl stuff seems not to have an effect. Have helped myself with
33 a rand(10), need to fix this back. So not checked in. Does the rand
36 The rand thing helps. The secondaryttl stuff was in the wrong line,
39 The pathdb stuff was because I called either _pathdb or __pathdb on the
40 wrong object. Fixed now.
42 * do not forget the dirtymark!
44 * It's not so beautiful if we never fetch the recentfiles that are not
45 the principal, even if this is correct behaviour. We really do not need
46 them after we have fetched the whole content.
48 OK, we want a switch for that: secondaryttl
50 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
53 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
55 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
56 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
57 it lets 40 follow. FIXED
59 * idea: have a new flag on recentfiles with the meaning: if this
60 changes, you're required to run a full rsync over all the files. The
61 reason why we set it would probably be: some foul happened. we injected
62 files in arbitrary places or didn't inject them although they changed.
63 The content of the flag? Timestamp? The relation between the
64 recentfiles would have to be inheritance from the principal, because any
65 out of band changes would soon later propagate to the next recentfile.
67 By upping the flag often one can easily ruin the slaves.
69 last out of band change? dirtymark?
71 Anyway, this implies that we read a potentially existing recentfile
74 And it implies that we have an eventloop that keeps us busy in 2-3
75 cycles, one for current stuff (tight loop) and one for the recentfiles
76 (cascade when principal has changed), one for the old stuff after a
79 * See that the long running process really only updates the principal
80 file unless it has missed a timespan during which something happened. If
81 nothing happened, it must notice even when it misses the timespan.
83 * fill up recentfiles with fake (historic) entries; fill up with
84 individual corrections; algorithm maybe to be done with bigfloat so that
85 we can always place something in the middle between two entries. Before
86 we must switch to bigfloat we could try to use Data::Float::nextup to
89 * we must throw away the pathdb when we have reached the end of Z. From
90 that moment we can have a very small pathdb because the only reason for
91 a pathdb is that we know to ignore old records in old files. We won't
92 need this pathdb again before the next full pass over the data is
93 necessary and then we will rebuild it as we go along.
95 * lookup by epoch and by path and use this ability on the pause to never
96 again register a file twice that doesn't need it.
98 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
100 * I think, Done::register_one is doing wrong in that it does not
101 conflate neighboring pieces. The covered() method cannot do this because
102 it has no recent_events array at hand. But register_one has it and could
103 do it and for some reason misses to do it (sometimes).
105 This means that the three tests I just wrote can probably not survive
106 because they test with an already broken Done structure.
108 The art now is to detect how it happens, then to reproduce, then write a
111 So from the logfile this is what happens: we have a good interval with
112 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
113 top of it. Locally we now mirror F2 and open a new done interval for it.
114 Then we mirror F1 but this time with the timestamp T1b. And when we then
115 try to close the gap, we do not find T1 but instead something older. We
116 should gladly accept this older piece and this would fix this bug.
120 * bug to fix: when the 1h file changes while rmirror is running, we do
121 correctly sync the new files but never switch to the 6h file but rather
122 stay in a rather quick loop that fetches the 1h file again and again.
124 Is it possible that we initialize a new object? Or does
125 get_remote_recentfile_as_tempfile overwrite something in myself?
127 Want a new option: _runstatusfile => $file which frequently dumps the
128 state of all recentfiles to a file.
130 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
132 * Todo: now teach update to verify the timestamp is about to write
133 against the previous and use _increase_a_bit if it doesn't comply with
134 strict monotony. DONE
136 * The problem of rounding. So far perl's default precision was
137 sufficient. One day it won't be. FakeFloat has an easy job when it is
138 only reading and other machines have written correctly. But when we want
139 to write a floating point number that is a bit larger than the other
140 one, then we need our own idea of precision.
142 Slaven said: just append a "1". This might be going towards the end of
143 usability too quickly. I'd like something that actually uses the decimal
144 system. Well, appending a 1 also does this but...
146 E.g. we have 1.0. nextup on this architecture is starting with
147 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
148 taken the 1.0000000000000003 and the next user comes and the time tells
149 him 1.0 again. He has to beat my number without stepping over the
150 nextup. This is much less space than I had when I chose 1,2,3.
152 What is also irritating is that nextup is architecture dependent. The
153 128 bit guy must choose very long numbers to fit in between whereas the
154 other one with 16 bit uses larger steps. But then the algorithm is the
155 same for both, so that would be a nice thing.
157 I see two situation where we need this. One is when Time::HiRes returns
158 us a value that is <= the last entry in our recentfile. In this case
159 (let's call it the end-case) we must fill the region between that number
160 and the next higher native floating point number. The other is when we
161 inject an old file into an old recentfile (we would then also set a new
162 dirtymark). We find the integer value already taken and need a slightly
163 different one (let's call it the middle-case). The difference between
164 the two situations is that the next user will want to find something
165 higher than my number in the end-case and something lower than my number
168 So I suggest we give the function both a value and an upper bound and it
169 calculates us a primitive middle. The upper bound in the middle-case is
170 the next integer. The upper bound on the end-case is the nextup floating
171 point number. But the latter poses another problem: if we have occupied
172 the middle m between x and nextup(x), then the nextup(m) will probably
173 not be the same as nextup(x) because some rounding will take place
174 before the nextup is calculated and when the rounding reaches the
175 nextup(x), we will end up at nextup(nextup(x)).
177 So we really need to consider the nextup and the nextdown from there and
178 then the middle and that's the number we may approach asymptotically.
181 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
183 * consider deprecating the use of RECENT.recent as a symlink. It turns
184 out to need extra hoops with the rsync options and just isn't worth it.
185 Or maybe these extra hoops are needed anyway for the rest of the tree?
186 Nope, can't be the case because not all filesystems support symlinks.
188 But before doing the large step, I'll deprecate the call of
189 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
190 only introduced to resolve RECENT.recent and complicates the routine far
191 beyond what it deserves.
193 DONE. Won't deprecate RECENT.recent, just moved its handling to the
196 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
198 * I think it's a bug that the rsync_option links must be set to true in
199 order to support RECENT.recent and that nobody cares to set it
200 automatically. Similar for ignore_link_stat_errors.
202 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
204 * Todo: find all todos together and make a plan what is missing for a
207 - verifytree or something like that. fsck maybe.
209 - rersyncrecent, the script itself
211 - a way to only mirror the recentfiles without mirroring the whole
212 remote system such that people can decide to mirror only partially see
213 also 2008-08-30. .shadow-xxx directory? this also needed for a
214 filesystem that is still incomplete and might need the mirrorfiles for
217 - long living objects that mirror again and again. Inject something
218 into ta, see how it goes over to tb.
220 - how do we continue filling up the DONE system when we use an object
221 for the second time? "fully covered" and "uptodate" or new terminology.
223 - overview called on the wrong file should be understandable
225 - the meta data field that must change when we fake something up so that
226 the downstream people know they have to re-fetch everything.
228 - how tolerant are we against missing files upstream? how do we keep
229 track? there are legitimate cases where we did read upstream index right
230 before a file got deleted there and then find that file as new and want
231 it. There are other cases that are not self healing and must be tracked
234 - how, exactly, do we have to deal with deletes? With rsync errors?
236 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
237 authors) failed: No such file or directory (2)
239 The file above is a delete in 1h and a new in file 1M and the
240 delete in the locally running rmirror did not get propagated to the 1M
241 object. Bug. And the consequence is a standstill.
243 It seems that a slave that works with a file below the principal needs
244 to merge things all the way up to get rid of later deletes. Or keep
245 track of all deletes and skip them later. So we need a trackdeletes.pm
246 similar to the done.pm?
248 see also 2008-08-20 about spurious deletes that really have no add
249 counterpart and yet they are not wrong.
251 - consider the effect when resyncing the recentfile takes longer than
252 the time per loop. Then we never rsync any file. We need to diagnose
253 that and force an increase of that loop time. But when we later are fast
254 enough again because the net has recovered, then we need to switch back
255 to original parameters. ERm, no, it's enough to keep syncing at least
256 one file before refetching an index file.
258 - remember to verify that no temp files are left lying around and the
261 - status file for not long running jobs that want to track upstream with
264 - revisit all XXX _float areas and study Sub::Exporter DONE
266 - persistent DB even though we just said we do not need it. Just for
267 extended capabilities and time savings when, for example, upstream
268 announces a reset and we get new recentfiles and could then limit
269 ourselves to a subset of files (those that have a changed epoch) in a
270 first pass and would only then do the loop to verify the rest. Or
273 * Todo: aggregate files should know their feed and finding the principal
274 should be done stepwise. (?)
276 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
277 left locks around. DONE
279 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
281 * maybe extend the _overview so that it always says if and where the
282 last file is in the next file and where the next event in the next rf
283 would lie. No, don't like this anymore. REJECT
285 * take the two new redundant tests out again, only the third must
288 * Todo: add a sanity check if the merged structure is really pointing to
289 a different rf and that this different rf is larger. DONE
291 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
293 * now test, if they are overlapping. And test if there is a file in the
294 next rf that would fit into this rf's interval.
296 1h 1222324012.8474 1222322541.7963 0.4086
297 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
298 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
299 1W 1222313218.3626 1221708477.5829 167.9835
301 I suspect that somebody writes a merged timestamp without having merged
302 and then somebody else relies on it.
304 If aggregate is running, the intervals must not be extravagated, if it
305 is not running, there must not be bounds, the total number of events in
306 the system must be counted and must be controlled throughout the tests.
307 That the test required the additional update was probably nonsense,
308 because aggregate can cut pieces too. FIXED & DONE
310 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
312 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
315 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
317 * Most apparent bug at the moment is that the recentfiles are fetched
318 too often. Only the principal should be fetched and if it has not
319 changed, the others should not be refetched. ATM I must admit that I'm
320 happy that we refetch more often than needed because I can more easily
321 fix bugs while the thing is running.
323 * Let's say, 1220474966.19501 is a timestamp of a file that is already
324 done but the done system does not know about it. The reason for the
325 failure is not known and we never reach the status uptodate because of
326 this. We must get over it.
328 Later it turns out that the origin server had a bug somewhere.
329 1220474966.19042 came after 1220474966.19501. Or better: it was in the
330 array of the recentfile one position above. The bug was my own.
332 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
334 * There is the race condition where the server does a delete and the
335 slave does not yet know and then tries to download it because he sees
336 the new. So for this time window we must be more tolerant against
337 failure. If we cannot download a file, we should just skip it and should
338 not retry immediately. The whole system should discover the lost thing
339 later. Keeping track with the DONE system should really be a no brainer.
341 But there is something more: the whole filesystem is a database and the
342 recentfiles are one possible representation of it. It's a pretty useful
343 representation I think that's why I have implemented something around
344 it. But for strictly local operation it has little value. For local
345 operation we would much rather have a database. So we would enter every
346 recentfile reading and every rsync operation and for every file the last
347 state change and what it leads to. Then we would always ignore older
348 records without the efforts involved with recentfiles.
350 The database would have: path,recentepoch,rsyncedon,deletedon
352 Oh well, not yet clear where this leads to.
354 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
356 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
357 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
359 * statusfile, maybe only on demand, alone to have a sharp debugging
360 tool. It is locked and all recentfiles dump themselves into it and we
361 can build a viewer that lets us know where we stand and what's inside.
363 * remember: only the principal recentfile needs expiration, all others
364 shall be expired by principal if it discovers that something has move
367 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
369 * Always check if we stringify to a higher value than in the entry
372 * And in covered make an additional check if we would be able to see a
373 numerical difference between the two numbers and if we can't then switch
374 to a different, more expensive algorithm. Do not want to be caught by
375 floating surprises. DONE
377 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
379 * caching has several aspects here: we can cache the interval of the
380 recentfile which only will change when the mtime of the file changes. We
381 must re-mirror the recentfile when its ttl has expired. Does have_read
382 tell you anything? It counts nothing at all. Only the mtime is
383 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
384 thing the have_mirrored because it is unrelated to the mtime.
386 * Robustness of floating point calculations! I always thought that the
387 string calculated by the origin server for the floating representation
388 of the epoch time is just a string. When we convert it to a number and
389 later back to a string, the other computer might come to a different
390 conclusion. This must not happen, we want to preserve it under any
391 circumstances. I will have to write tests with overlong sequences that
392 get lost in arithmetic and must see if all still works well. DONE
394 But one fragile point remains: if one host considers a>b and the other
395 one considers them == but no eq. To prevent this, we must probably do
396 some extra homework. DONE
398 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
400 * the concept of tracking DONE needs an object per recentfile that has
401 something like these methods:
403 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
405 covered() register() covered()
407 The unclear thing is how we translate points in time into intervals. We
408 could pass a reference to the current recent_events array when running
409 we_have(xxx) and let the DONE object iterate over it such that it only
410 has to store a list of intervals that can melt into each other. Ah, even
411 passing the list together with a list of indexes seems feasiable.
413 Or maybe ask for the inverted list?
415 Whenever the complete array is covered by the interval we say we are
416 fully covered and if the recentfile is not expired, we are uptodate.
418 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
420 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
422 * need a way to "return" the next entry after the end of a list. When
423 the caller says "before" or "after" we would like to know if he could
424 cover that interval/threshold or not because this influences the effect
425 of a newer timestamp of that recentfile. DONE with $opt{info}.
427 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
429 * one of the next things to tackle: the equivalent of csync2 -TIXU.
431 loop implies tixu (?). Nope, something like --statefile decides. Per
434 T test, I init, X including removals, U nodirtymark
436 So we have no concept of dirtymarks, we only trust that since we are
437 running we have observed everything steadily. But people will not let
438 this program run forever so we must consider both startup penalty and
439 book keeping for later runs. We keep this for later. For now we write a
440 long running mirror that merges several intervals.
442 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
444 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
447 * rersyncrecent, the script: default to one week. The name of the switch
448 is --after. Other switches? --loop!
450 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
452 * need a switch --skip-deletes (?)
454 * need a switch --enduser that tells us that the whole tempfile
455 discipline is not needed when there is no downstream user. (?)
457 Without this switch we cannot have a reasonable recent.pl that just
458 displays the recent additions. Either we accept to download everything.
459 Or we download temporary files without the typical rsync protocol
462 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
463 File::Temp::tempdir, this might be a win.
465 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
467 * apropos missing: we have no push, we never know the downstream
468 servers. People who know their downstream hosts and want to ascertain
469 something will want additional methods we have never thought about, like
470 update or delete a certain file.
472 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
474 * tempted to refactor rmirror into resolve_symlink, localize, etc.
475 Curious if rsync_options=links equal 0 vs. 1 will make the expected
478 * rsync options: it's a bit of a pain that we usually need several rsync
479 options, like compress, links, times, checksum and that there is no
480 reasonable default except the original rsync default. I think wee can
481 safely assume that the rsync options are shared between all recentfile
482 instances within one recent tree.
484 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
486 * deletes: if a delete follows an add quickly enough it may happen that
487 a downstream mirror did not see the add at all! It seems this needs to
488 be mentioned somewhere. The point here is that even if the downstream is
489 never missing the principal timeframe it may encounter a "delete" that
490 has no complimentary "add" anywhere.
492 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
494 * I suspect the treat of metadata is incorrect during read or something.
495 The bug that I am watching is that between 06:08 and 06:09 the 6h file
496 contained more than 6 hours worth of data. At 06:08 we merged into the
497 1d file. We need to take snapshots of the 6h file over the course of an
498 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
501 Much worse: watching the 1h file: right at the moment (at 06:35) it
502 covers 1218867584-1219120397 which is 70 hours.
504 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
505 08:19:44 2008, that is when I checked out last time, so it seems to be
506 aggregating and never truncating?
508 No, correct is: it is never truncating; but wrong is: it is aggregating.
509 It does receive a lot of events from time to time from a larger file.
510 Somehow a large file gets merged into the small one and because the
511 "meta/merged" attribute is missing, nobody is paying attention. I
512 believe that I can fix this by making sure that metadata are honoured
513 during read. DONE and test adjusted.
515 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
517 * grand renaming plan
519 remotebase => remoteroot to fit well with localroot DONE
520 local_path() => localroot seems to me should already work DONE
521 recentfile_basename => rfilename no need to stress it has no slash DONE
523 filenameroot??? Doesn't seem too bad to me today. Maybe something like
524 kern? It would anyway need a deprecation cycle because it is an
525 important constructor.
527 * I like the portability that Data::Serializer brings us but the price
528 is that some day we might find out that it is slowing us a bit. We'll
531 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
533 * should we not enter the interval of the principal (or the interval of
534 the merging file?) in every aggregated/merged file?
536 * we should aim at a first release and give up on thinking about
537 sanitizing stuff and zloop. Let's just admit that a full traditional
538 rsync is the only available sanitizer ATM. Otherwise it's complicated
539 stuff: sanitizing on the origin server, sanitizing on the slaves,
540 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
541 get the basics out before this becomes a major cause for mess.
543 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
545 * On OSes not supporting symlinks we expect that RECENT.recent contains
546 the contents of the principal recentfile. Actually this is identical on
547 systems supporting symlinks. Simple, what follows from that is that we
548 need to keep the serializer in the metadata because we cannot read it
549 from the filename, doesn't it? Of course not. It's a chicken and egg
550 problem. This leaves us with the problem to actually parse the
551 serialized data to find out in which format it is. So who can do the 4
552 or 5 magics we wanted to support? File::LibMagic?
554 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
556 * remotebase and recentfile_basename are ugly names. Now that we need a
557 word for the shortest/principal/driving recentfile too we should do
560 localroot is good. rfile is good. local_path() is bad, local_path($path)
561 is medium, filenameroot() is bad, remotebase is bad, recentfile is
564 Up to now remotebase was the string that described the remote root
565 directory in rsync notation, like pause.perl.org::authors. And
566 recentfile_basename was "RECENT-1h.yaml".
568 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
570 * The test that was added in today's checkin is a good start for a test
571 of rmirror. We should have more methods in Recent.pm: verify,
572 addmissingfiles. We should verify the current tree, then rmirror it and
573 then verifytree the copy. We could then add some arbitrary file and let
574 it be discovered by addmissingfiles, then rmirror again and then
575 verifytree the copy again.
577 Then we could start stealing from csync2 sqlite database [no port to
578 OSX!] and fill a local DB. And methods to compare the database with the
579 recentfiles. Our strength is that in principle we could maintain state
580 with a single float. We have synced up to 1234567890.123456. If the Z
581 file does not add new files all we have to do is mirror the new ones and
584 This makes it clear that we should extend current protocol and declare
585 that we cheat when we add files too late, just to help the other end
586 keeping track. Ah yes, that's what was meant when zloop was mentioned
589 Maybe need to revisit File::Mirror to help me with this task.
591 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
593 * There must be an allow-me-to-truncate flag in every recentfile.
594 Without it one could construct a sequence of updates winning the locking
595 battle against the aggregator. Only if an aggregator has managed to
596 merge data over to the next level, truncating can be allowed. DONE with
599 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
601 * We should probably guarantee that no duplicates enter the aggregator
604 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
606 * To get merge operation faster would need a good benchmark test. What
607 02 spits out isn't reliable enough and is dominated by many other
610 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
611 Author: Andreas J. Koenig <k@k75.(none)>
612 Date: Sat Aug 2 07:58:04 2008 +0200
616 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
617 Author: Andreas J. Koenig <k@k75.(none)>
618 Date: Sat Aug 2 11:40:29 2008 +0200
620 there was a lot of trying but the effect is hardly measurable with
623 * overhead of connecting seems high. When setting
624 max_files_per_connection to 1 we see that.
626 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
628 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
630 25138.2215600014/3600 = 6.98283932222261
632 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
633 be a bug in the aggregator. FIXED
635 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
637 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
638 we have it, can we mirror it, mirror it!
640 I fear this needs a new class which might be called
641 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
642 and treat them as an entity. I realize that a single recentfile may be
643 sufficient for certain tasks and that it is handy for the low level
644 programmer but it is not nice to use. If there is a delete in the 1h
645 file then the 6h file still contains it. Seekers of the best information
646 need to combine at least some of the recentfiles most of the time.
648 There is the place for the Z loop!
650 But the combination is something to collect in a database, isn't it. Did
651 csync2 just harrumph?
653 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
655 * it just occurred to me that hosts in the same mirroring pool could
656 help out each other even without rewriting the recentfile. Just fetch
657 the stuff to mirror from several places, bingo. But that's something
658 that should rather live in a separate package or in rsync directly.
660 * cronjobs are unsuited because with ntp they would all come at the full
661 minute and disturb each other. Besides that I'd hate to have a backbone
662 with more than a few seconds latency.
664 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
666 * a second rsync server with access control for PAUSE. Port? 873 is the
667 standard port, let's take 8873.
669 * if there were a filesystem based on this, it would have a slow access
670 to inexistent files. It would probably provide wrong readdir (only based
671 on current content) or also a slow one (based on a recentfile written
672 after the call). But it would provide fast access to existing files. Or
673 one would deliberately allow slightly blurred answers based on some
674 sqlite reflection of the recentfiles.
676 * todo: write a variant of mirror() that combines two or more
677 recentfiles and treats them like one
679 * todo: signal handler to remove the tempfile
681 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
683 * now that we have the symlink I forgot how it should be used in
686 * the z loop: add missing files to Z file. Just append them (instead of
687 prepending). So one guy prepends something from the Y file from time to
688 time and another guy appends something rather frequently. Collecting
689 pond. When Y merges into Z, things get epoch and the collecting pond
690 gets smaller. What exactly are "missing files"?
692 take note of current epoch of the alpha file, let's call it the
695 find all files on disk
697 remove all files registered in the recentworld up to recent-ts
699 remove all files that have been deleted after recent-ts according to
702 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
704 * rersyncrecent might be a cronjob with a (locked) state file which
705 contains things like after and maybe last z sync or such?
707 rrr-mirror might be an alternative name but how would we justify the
708 three Rs when there is no Re-Rsync-Recent?
710 With the --loop parameter it is an endless loop, without it is no loop.
711 At least this is simple.
713 * todo: new accssor z-interval specifies how often the Z file is updated
714 against the filesystem. We probably want no epoch stamp on these
715 entries. And we want to be able to filter the entries (e.g. no
716 by-modules and by-category tree)
718 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
720 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
721 into Z. We do not need the other files filled up with filesystem
724 * need interface to query for a file in order to NOT call update on
725 PAUSE a second time within a short time.
727 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
729 * recommended update interval? Makes no sense, is different for
736 change-log-default-name: "Todo"