1 2008-10-11 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * another bug is the fact that the mirror command deletes files before
4 it unhides the index file, thus confusing downstream slaves. We must not
5 delete before unhiding and must delete after unhiding. FIXED.
7 * new complication about the slave that is playing a server role.
8 Currently we mirror from newest to oldest with a hidden temporary file
9 as index. And when one file is finished, we unhide the index file.
10 Imagine the cascading server/slave is dead for a day. It then starts
11 mirroring again with the freshest thing and unhides the freshest index
12 file when it has worked through it. In that moment it exposes a time
13 hole. Because it now works on the second recentfile which is still
16 We currently do nothing special to converge after such a drop out. At
17 least not intentionally and robustly and thought through.
19 The algorithm we use to seed the next file needs quite a lot of more
20 robustness than it currently has. Something to do with looking at the
21 merged element of the next rf and when it has dropped off, we seed
22 immediately. And if it ramains dropped off, we seed again, of course.
24 Nope, looking from smaller to larger RFS we look at the merged element
25 of this RF and at the minmax/max element of the next RF. If that
26 $rf[next]->{minmax}{max} >= $rf[this]->{merged}{epoch}, then we can stop
29 And we need a public accessor seed and unseed or seeded. But not the mix
30 of public and private stuff that then is used behind the back.
32 And then the secondary* stuff must go.
34 And we must understand what the impact is on the DONE system. Can it go
35 unnoticed that there was a hole? And could the DONE system have decided
36 the hole is covered? This sould be testable with three directories where
37 the middle stops working for a while.
39 * dirtymark now lives in Recentfile, needs to be used in rmirror.
41 2008-10-10 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
43 * Slaven suggests to have the current epoch or the whole current
44 recentfile available from the HTTP server and take it away with
45 keepalive. This direction goes the granularity down to subseconds.
47 We might want to rewrite everything to factor out transport and allow
48 the whole thing to run via HTTP.
50 2008-10-09 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
52 * are we sure we do NOT LEAVE DOT FILES around? Especially on the
55 * smoker on k81 fetching from k75 to verify cascading works. See
56 2008-07-17 in upgradexxx and rsync-over-recentfile-3.pl.
58 * maybe the loop should wait for CHECKSUMS file after every upload. And
59 CPAN.pm needs to deal with timestamps in the future.
61 * do not forget the dirtymark!
63 Text: have a new flag on recentfiles with the meaning: if this
64 changes, you're required to run a full rsync over all the files. The
65 reason why we set it would probably be: some foul happened. we injected
66 files in arbitrary places or didn't inject them although they changed.
67 The content of the flag? Timestamp? The relation between the
68 recentfiles would have to be inheritance from the principal, because any
69 out of band changes would soon later propagate to the next recentfile.
71 By upping the flag often one can easily ruin the slaves.
73 last out of band change? dirtymark?
75 Anyway, this implies that we read a potentially existing recentfile
78 And it implies that we have an eventloop that keeps us busy in 2-3
79 cycles, one for current stuff (tight loop) and one for the recentfiles
80 (cascade when principal has changed), one for the old stuff after a
83 And it implies that the out-of-band change in any of the recentfiles
84 must have a lock on the principal file and there is the place to set the
87 * after the dirtymark is done: fill up recentfiles with fake (historic)
88 entries; fill up with individual corrections; algorithm maybe to be done
89 with bigfloat so that we can always place something in the middle
90 between two entries. Before we must switch to bigfloat we could try to
91 use Data::Float::nextup to get the.
93 * lookup by epoch and by path and use this ability on the pause to never
94 again register a file twice that doesn't need it.
96 * Inotify2 on an arbitrary tree and then play with that instead of PAUSE
99 * start a FAQ, especially quick start guide questions. Also to aid those
100 problematic areas where we have no good solution, like the "links"
103 * wish feedback when we are slow.
107 * Remove a few DEBUG statements.
109 * The multiple-rrr way of doing things needs a new option to rmirror,
110 like piecemeal or so. Not urgent because after the first pass through,
111 things run smoothely. It's only ugly during the first pass.
113 * I have the suspicion that the code is broken that decides if the
114 neighboring RF needs to be seeded. I fear when too much time has gone
115 between two calls (in our case more than one hour), it would not seed
116 the neighbor. Of course this will never be noticed, so we need a good
119 * local/localroot confusion: I currently pass both options but one must
122 * accounts for early birds on PAUSE rsync daemon.
124 * hardcoded 20 seconds
126 * who mirrors the index? DOING now.
128 * which CPAN mirrors offer rsync?
130 * visit all XXX, visit all _float places
132 * rename the pathdb stuff, it's too confusing. No idea how.
134 * rrr-inotify, backpan, rrr-register
136 2008-10-08 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
138 * current bugs: the pathdb seems to get no reset, the seeding of the
139 secondaryttl stuff seems not to have an effect. Have helped myself with
140 a rand(10), need to fix this back. So not checked in. Does the rand
143 The rand thing helps. The secondaryttl stuff was in the wrong line,
146 The pathdb stuff was because I called either _pathdb or __pathdb on the
147 wrong object. FIXED now.
149 * It's not so beautiful if we never fetch the recentfiles that are not
150 the principal, even if this is correct behaviour. We really do not need
151 them after we have fetched the whole content.
153 OK, we want a switch for that: secondaryttl DONE
155 2008-10-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
157 * bug: rrr-news --max does not count correctly. with "35" it shows me 35
158 lines but with 36 it shows 110. First it repeats 35, gives 70, and then
159 it lets 40 follow. FIXED
161 * See that the long running process really only updates the principal
162 file unless it has missed a timespan during which something happened. If
163 nothing happened, it must notice even when it misses the timespan. DONE
165 * we must throw away the pathdb when we have reached the end of Z. From
166 that moment we can have a very small pathdb because the only reason for
167 a pathdb is that we know to ignore old records in old files. We won't
168 need this pathdb again before the next full pass over the data is
169 necessary and then we will rebuild it as we go along. DONE
171 2008-10-06 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
173 * I think, Done::register_one is doing wrong in that it does not
174 conflate neighboring pieces. The covered() method cannot do this because
175 it has no recent_events array at hand. But register_one has it and could
176 do it and for some reason misses to do it (sometimes).
178 This means that the three tests I just wrote can probably not survive
179 because they test with an already broken Done structure.
181 The art now is to detect how it happens, then to reproduce, then write a
184 So from the logfile this is what happens: we have a good interval with
185 newest file being F1 at T1. Now remotely F1 gets a change and F2 goes on
186 top of it. Locally we now mirror F2 and open a new done interval for it.
187 Then we mirror F1 but this time with the timestamp T1b. And when we then
188 try to close the gap, we do not find T1 but instead something older. We
189 should gladly accept this older piece and this would fix this bug.
193 * bug to fix: when the 1h file changes while rmirror is running, we do
194 correctly sync the new files but never switch to the 6h file but rather
195 stay in a rather quick loop that fetches the 1h file again and again.
197 Is it possible that we initialize a new object? Or does
198 get_remote_recentfile_as_tempfile overwrite something in myself?
200 Want a new option: _runstatusfile => $file which frequently dumps the
201 state of all recentfiles to a file.
205 2008-10-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
207 * Todo: now teach update to verify the timestamp is about to write
208 against the previous and use _increase_a_bit if it doesn't comply with
209 strict monotony. DONE
211 * The problem of rounding. So far perl's default precision was
212 sufficient. One day it won't be. FakeFloat has an easy job when it is
213 only reading and other machines have written correctly. But when we want
214 to write a floating point number that is a bit larger than the other
215 one, then we need our own idea of precision.
217 Slaven said: just append a "1". This might be going towards the end of
218 usability too quickly. I'd like something that actually uses the decimal
219 system. Well, appending a 1 also does this but...
221 E.g. we have 1.0. nextup on this architecture is starting with
222 1.0000000000000004. So there is a gap to fill: 1,2,3. Now I have
223 taken the 1.0000000000000003 and the next user comes and the time tells
224 him 1.0 again. He has to beat my number without stepping over the
225 nextup. This is much less space than I had when I chose 1,2,3.
227 What is also irritating is that nextup is architecture dependent. The
228 128 bit guy must choose very long numbers to fit in between whereas the
229 other one with 16 bit uses larger steps. But then the algorithm is the
230 same for both, so that would be a nice thing.
232 I see two situation where we need this. One is when Time::HiRes returns
233 us a value that is <= the last entry in our recentfile. In this case
234 (let's call it the end-case) we must fill the region between that number
235 and the next higher native floating point number. The other is when we
236 inject an old file into an old recentfile (we would then also set a new
237 dirtymark). We find the integer value already taken and need a slightly
238 different one (let's call it the middle-case). The difference between
239 the two situations is that the next user will want to find something
240 higher than my number in the end-case and something lower than my number
243 So I suggest we give the function both a value and an upper bound and it
244 calculates us a primitive middle. The upper bound in the middle-case is
245 the next integer. The upper bound on the end-case is the nextup floating
246 point number. But the latter poses another problem: if we have occupied
247 the middle m between x and nextup(x), then the nextup(m) will probably
248 not be the same as nextup(x) because some rounding will take place
249 before the nextup is calculated and when the rounding reaches the
250 nextup(x), we will end up at nextup(nextup(x)).
252 So we really need to consider the nextup and the nextdown from there and
253 then the middle and that's the number we may approach asymptotically.
256 2008-10-03 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
258 * consider deprecating the use of RECENT.recent as a symlink. It turns
259 out to need extra hoops with the rsync options and just isn't worth it.
260 Or maybe these extra hoops are needed anyway for the rest of the tree?
261 Nope, can't be the case because not all filesystems support symlinks.
263 But before doing the large step, I'll deprecate the call of
264 get_remote_recentfile_as_tempfile with an argument. Rememberr this was
265 only introduced to resolve RECENT.recent and complicates the routine far
266 beyond what it deserves.
268 DONE. Won't deprecate RECENT.recent, just moved its handling to the
271 2008-10-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
273 * I think it's a bug that the rsync_option links must be set to true in
274 order to support RECENT.recent and that nobody cares to set it
275 automatically. Similar for ignore_link_stat_errors.
277 2008-09-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
279 * Todo: find all todos together and make a plan what is missing for a
282 - verifytree or something like that. fsck maybe.
284 - rersyncrecent, the script itself? What it do?
286 - a way to only mirror the recentfiles without mirroring the whole
287 remote system such that people can decide to mirror only partially see
288 also 2008-08-30. .shadow-xxx directory? this also needed for a
289 filesystem that is still incomplete and might need the mirrorfiles for
292 - long living objects that mirror again and again. Inject something
293 into ta, see how it goes over to tb.
295 - how do we continue filling up the DONE system when we use an object
296 for the second time? "fully covered" and "uptodate" or new terminology.
298 - overview called on the wrong file should be understandable
300 - the meta data field that must change when we fake something up so that
301 the downstream people know they have to re-fetch everything.
303 - how tolerant are we against missing files upstream? how do we keep
304 track? there are legitimate cases where we did read upstream index right
305 before a file got deleted there and then find that file as new and want
306 it. There are other cases that are not self healing and must be tracked
309 - how, exactly, do we have to deal with deletes? With rsync errors?
311 rsync: link_stat "/id/K/KA/KARMAN/Rose-HTMLx-Form-Related-0.07.meta" (in
312 authors) failed: No such file or directory (2)
314 The file above is a delete in 1h and a new in file 1M and the
315 delete in the locally running rmirror did not get propagated to the 1M
316 object. Bug. And the consequence is a standstill.
318 It seems that a slave that works with a file below the principal needs
319 to merge things all the way up to get rid of later deletes. Or keep
320 track of all deletes and skip them later. So we need a trackdeletes.pm
321 similar to the done.pm?
323 see also 2008-08-20 about spurious deletes that really have no add
324 counterpart and yet they are not wrong.
326 - consider the effect when resyncing the recentfile takes longer than
327 the time per loop. Then we never rsync any file. We need to diagnose
328 that and force an increase of that loop time. But when we later are fast
329 enough again because the net has recovered, then we need to switch back
330 to original parameters. ERm, no, it's enough to keep syncing at least
331 one file before refetching an index file.
333 - remember to verify that no temp files are left lying around and the
336 - status file for not long running jobs that want to track upstream with
339 - revisit all XXX _float areas and study Sub::Exporter DONE
341 - persistent DB even though we just said we do not need it. Just for
342 extended capabilities and time savings when, for example, upstream
343 announces a reset and we get new recentfiles and could then limit
344 ourselves to a subset of files (those that have a changed epoch) in a
345 first pass and would only then do the loop to verify the rest. Or
348 * Todo: aggregate files should know their feed and finding the principal
349 should be done stepwise. (?)
351 * Todo: DESTROY thing that unlocks. Today when I left the debuggerr I
352 left locks around. DONE
354 2008-09-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
356 * maybe extend the _overview so that it always says if and where the
357 last file is in the next file and where the next event in the next rf
358 would lie. No, don't like this anymore. REJECT
360 * take the two new redundant tests out again, only the third must
363 * Todo: add a sanity check if the merged structure is really pointing to
364 a different rf and that this different rf is larger. DONE
366 2008-09-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
368 * now test, if they are overlapping. And test if there is a file in the
369 next rf that would fit into this rf's interval.
371 1h 1222324012.8474 1222322541.7963 0.4086
372 6h 1222320411.2760 1222304207.6931 4.5010 missing overlap/gap!
373 1d 1222320411.2760 1222238750.5071 22.6835 large overlap
374 1W 1222313218.3626 1221708477.5829 167.9835
376 I suspect that somebody writes a merged timestamp without having merged
377 and then somebody else relies on it.
379 If aggregate is running, the intervals must not be extravagated, if it
380 is not running, there must not be bounds, the total number of events in
381 the system must be counted and must be controlled throughout the tests.
382 That the test required the additional update was probably nonsense,
383 because aggregate can cut pieces too. FIXED & DONE
385 2008-09-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
387 * rrr-aggregate seems to rewrite the RECENT file even if nothing has
390 2008-09-21 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
392 * Most apparent bug at the moment is that the recentfiles are fetched
393 too often. Only the principal should be fetched and if it has not
394 changed, the others should not be refetched. ATM I must admit that I'm
395 happy that we refetch more often than needed because I can more easily
396 fix bugs while the thing is running.
398 * Let's say, 1220474966.19501 is a timestamp of a file that is already
399 done but the done system does not know about it. The reason for the
400 failure is not known and we never reach the status uptodate because of
401 this. We must get over it.
403 Later it turns out that the origin server had a bug somewhere.
404 1220474966.19042 came after 1220474966.19501. Or better: it was in the
405 array of the recentfile one position above. The bug was my own.
407 2008-09-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
409 * There is the race condition where the server does a delete and the
410 slave does not yet know and then tries to download it because he sees
411 the new. So for this time window we must be more tolerant against
412 failure. If we cannot download a file, we should just skip it and should
413 not retry immediately. The whole system should discover the lost thing
414 later. Keeping track with the DONE system should really be a no brainer.
416 But there is something more: the whole filesystem is a database and the
417 recentfiles are one possible representation of it. It's a pretty useful
418 representation I think that's why I have implemented something around
419 it. But for strictly local operation it has little value. For local
420 operation we would much rather have a database. So we would enter every
421 recentfile reading and every rsync operation and for every file the last
422 state change and what it leads to. Then we would always ignore older
423 records without the efforts involved with recentfiles.
425 The database would have: path,recentepoch,rsyncedon,deletedon
427 Oh well, not yet clear where this leads to.
429 2008-09-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
431 * Bug: the bigloop ran into a funny endless loop after EWILHELM uploaded
432 Module-Build. It *only* rsynced the "1h" recentfile from that moment on.
434 * statusfile, maybe only on demand, alone to have a sharp debugging
435 tool. It is locked and all recentfiles dump themselves into it and we
436 can build a viewer that lets us know where we stand and what's inside.
438 * remember: only the principal recentfile needs expiration, all others
439 shall be expired by principal if it discovers that something has move
442 2008-09-18 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
444 * Always check if we stringify to a higher value than in the entry
447 * And in covered make an additional check if we would be able to see a
448 numerical difference between the two numbers and if we can't then switch
449 to a different, more expensive algorithm. Do not want to be caught by
450 floating surprises. DONE
452 2008-09-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
454 * caching has several aspects here: we can cache the interval of the
455 recentfile which only will change when the mtime of the file changes. We
456 must re-mirror the recentfile when its ttl has expired. Does have_read
457 tell you anything? It counts nothing at all. Only the mtime is
458 interesting. The ntuple mtime, low-epoch, high-epoch. And as a separate
459 thing the have_mirrored because it is unrelated to the mtime.
461 * Robustness of floating point calculations! I always thought that the
462 string calculated by the origin server for the floating representation
463 of the epoch time is just a string. When we convert it to a number and
464 later back to a string, the other computer might come to a different
465 conclusion. This must not happen, we want to preserve it under any
466 circumstances. I will have to write tests with overlong sequences that
467 get lost in arithmetic and must see if all still works well. DONE
469 But one fragile point remains: if one host considers a>b and the other
470 one considers them == but no eq. To prevent this, we must probably do
471 some extra homework. DONE
473 2008-09-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
475 * the concept of tracking DONE needs an object per recentfile that has
476 something like these methods:
478 do_we_have(xxx), we_have(xxx), do_we_have_all(xxx,yyy), reset()
480 covered() register() covered()
482 The unclear thing is how we translate points in time into intervals. We
483 could pass a reference to the current recent_events array when running
484 we_have(xxx) and let the DONE object iterate over it such that it only
485 has to store a list of intervals that can melt into each other. Ah, even
486 passing the list together with a list of indexes seems feasiable.
488 Or maybe ask for the inverted list?
490 Whenever the complete array is covered by the interval we say we are
491 fully covered and if the recentfile is not expired, we are uptodate.
493 2008-09-07 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
495 2008-09-05 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
497 * need a way to "return" the next entry after the end of a list. When
498 the caller says "before" or "after" we would like to know if he could
499 cover that interval/threshold or not because this influences the effect
500 of a newer timestamp of that recentfile. DONE with $opt{info}.
502 2008-09-04 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
504 * one of the next things to tackle: the equivalent of csync2 -TIXU.
506 loop implies tixu (?). Nope, something like --statefile decides. Per
509 T test, I init, X including removals, U nodirtymark
511 So we have no concept of dirtymarks, we only trust that since we are
512 running we have observed everything steadily. But people will not let
513 this program run forever so we must consider both startup penalty and
514 book keeping for later runs. We keep this for later. For now we write a
515 long running mirror that merges several intervals.
517 2008-09-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
519 * need to speed up the 02 test, it's not clever to sleep so much. Reduce
522 * rersyncrecent, the script: default to one week. The name of the switch
523 is --after. Other switches? --loop!
525 2008-08-30 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
527 * need a switch --skip-deletes (?)
529 * need a switch --enduser that tells us that the whole tempfile
530 discipline is not needed when there is no downstream user. (?)
532 Without this switch we cannot have a reasonable recent.pl that just
533 displays the recent additions. Either we accept to download everything.
534 Or we download temporary files without the typical rsync protocol
537 Or maybe the switch is --tmpdir? If --tmpdir would mean: do not use
538 File::Temp::tempdir, this might be a win.
540 2008-08-29 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
542 * apropos missing: we have no push, we never know the downstream
543 servers. People who know their downstream hosts and want to ascertain
544 something will want additional methods we have never thought about, like
545 update or delete a certain file.
547 2008-08-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
549 * tempted to refactor rmirror into resolve_symlink, localize, etc.
550 Curious if rsync_options=links equal 0 vs. 1 will make the expected
553 * rsync options: it's a bit of a pain that we usually need several rsync
554 options, like compress, links, times, checksum and that there is no
555 reasonable default except the original rsync default. I think wee can
556 safely assume that the rsync options are shared between all recentfile
557 instances within one recent tree.
559 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
561 * deletes: if a delete follows an add quickly enough it may happen that
562 a downstream mirror did not see the add at all! It seems this needs to
563 be mentioned somewhere. The point here is that even if the downstream is
564 never missing the principal timeframe it may encounter a "delete" that
565 has no complimentary "add" anywhere.
567 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
569 * I suspect the treat of metadata is incorrect during read or something.
570 The bug that I am watching is that between 06:08 and 06:09 the 6h file
571 contained more than 6 hours worth of data. At 06:08 we merged into the
572 1d file. We need to take snapshots of the 6h file over the course of an
573 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
576 Much worse: watching the 1h file: right at the moment (at 06:35) it
577 covers 1218867584-1219120397 which is 70 hours.
579 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
580 08:19:44 2008, that is when I checked out last time, so it seems to be
581 aggregating and never truncating?
583 No, correct is: it is never truncating; but wrong is: it is aggregating.
584 It does receive a lot of events from time to time from a larger file.
585 Somehow a large file gets merged into the small one and because the
586 "meta/merged" attribute is missing, nobody is paying attention. I
587 believe that I can fix this by making sure that metadata are honoured
588 during read. DONE and test adjusted.
590 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
592 * grand renaming plan
594 remotebase => remoteroot to fit well with localroot DONE
595 local_path() => localroot seems to me should already work DONE
596 recentfile_basename => rfilename no need to stress it has no slash DONE
598 filenameroot??? Doesn't seem too bad to me today. Maybe something like
599 kern? It would anyway need a deprecation cycle because it is an
600 important constructor.
602 * I like the portability that Data::Serializer brings us but the price
603 is that some day we might find out that it is slowing us a bit. We'll
606 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
608 * should we not enter the interval of the principal (or the interval of
609 the merging file?) in every aggregated/merged file?
611 * we should aim at a first release and give up on thinking about
612 sanitizing stuff and zloop. Let's just admit that a full traditional
613 rsync is the only available sanitizer ATM. Otherwise it's complicated
614 stuff: sanitizing on the origin server, sanitizing on the slaves,
615 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
616 get the basics out before this becomes a major cause for mess.
618 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
620 * On OSes not supporting symlinks we expect that RECENT.recent contains
621 the contents of the principal recentfile. Actually this is identical on
622 systems supporting symlinks. Simple, what follows from that is that we
623 need to keep the serializer in the metadata because we cannot read it
624 from the filename, doesn't it? Of course not. It's a chicken and egg
625 problem. This leaves us with the problem to actually parse the
626 serialized data to find out in which format it is. So who can do the 4
627 or 5 magics we wanted to support? File::LibMagic?
629 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
631 * remotebase and recentfile_basename are ugly names. Now that we need a
632 word for the shortest/principal/driving recentfile too we should do
635 localroot is good. rfile is good. local_path() is bad, local_path($path)
636 is medium, filenameroot() is bad, remotebase is bad, recentfile is
639 Up to now remotebase was the string that described the remote root
640 directory in rsync notation, like pause.perl.org::authors. And
641 recentfile_basename was "RECENT-1h.yaml".
643 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
645 * The test that was added in today's checkin is a good start for a test
646 of rmirror. We should have more methods in Recent.pm: verify,
647 addmissingfiles. We should verify the current tree, then rmirror it and
648 then verifytree the copy. We could then add some arbitrary file and let
649 it be discovered by addmissingfiles, then rmirror again and then
650 verifytree the copy again.
652 Then we could start stealing from csync2 sqlite database [no port to
653 OSX!] and fill a local DB. And methods to compare the database with the
654 recentfiles. Our strength is that in principle we could maintain state
655 with a single float. We have synced up to 1234567890.123456. If the Z
656 file does not add new files all we have to do is mirror the new ones and
659 This makes it clear that we should extend current protocol and declare
660 that we cheat when we add files too late, just to help the other end
661 keeping track. Ah yes, that's what was meant when zloop was mentioned
664 Maybe need to revisit File::Mirror to help me with this task.
666 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
668 * There must be an allow-me-to-truncate flag in every recentfile.
669 Without it one could construct a sequence of updates winning the locking
670 battle against the aggregator. Only if an aggregator has managed to
671 merge data over to the next level, truncating can be allowed. DONE with
674 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
676 * We should probably guarantee that no duplicates enter the aggregator
679 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
681 * To get merge operation faster would need a good benchmark test. What
682 02 spits out isn't reliable enough and is dominated by many other
685 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
686 Author: Andreas J. Koenig <k@k75.(none)>
687 Date: Sat Aug 2 07:58:04 2008 +0200
691 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
692 Author: Andreas J. Koenig <k@k75.(none)>
693 Date: Sat Aug 2 11:40:29 2008 +0200
695 there was a lot of trying but the effect is hardly measurable with
698 * overhead of connecting seems high. When setting
699 max_files_per_connection to 1 we see that.
701 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
703 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
705 25138.2215600014/3600 = 6.98283932222261
707 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
708 be a bug in the aggregator. FIXED
710 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
712 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
713 we have it, can we mirror it, mirror it!
715 I fear this needs a new class which might be called
716 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
717 and treat them as an entity. I realize that a single recentfile may be
718 sufficient for certain tasks and that it is handy for the low level
719 programmer but it is not nice to use. If there is a delete in the 1h
720 file then the 6h file still contains it. Seekers of the best information
721 need to combine at least some of the recentfiles most of the time.
723 There is the place for the Z loop!
725 But the combination is something to collect in a database, isn't it. Did
726 csync2 just harrumph?
728 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
730 * it just occurred to me that hosts in the same mirroring pool could
731 help out each other even without rewriting the recentfile. Just fetch
732 the stuff to mirror from several places, bingo. But that's something
733 that should rather live in a separate package or in rsync directly.
735 * cronjobs are unsuited because with ntp they would all come at the full
736 minute and disturb each other. Besides that I'd hate to have a backbone
737 with more than a few seconds latency.
739 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
741 * a second rsync server with access control for PAUSE. Port? 873 is the
742 standard port, let's take 8873.
744 * if there were a filesystem based on this, it would have a slow access
745 to inexistent files. It would probably provide wrong readdir (only based
746 on current content) or also a slow one (based on a recentfile written
747 after the call). But it would provide fast access to existing files. Or
748 one would deliberately allow slightly blurred answers based on some
749 sqlite reflection of the recentfiles.
751 * todo: write a variant of mirror() that combines two or more
752 recentfiles and treats them like one
754 * todo: signal handler to remove the tempfile
756 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
758 * now that we have the symlink I forgot how it should be used in
761 * the z loop: add missing files to Z file. Just append them (instead of
762 prepending). So one guy prepends something from the Y file from time to
763 time and another guy appends something rather frequently. Collecting
764 pond. When Y merges into Z, things get epoch and the collecting pond
765 gets smaller. What exactly are "missing files"?
767 take note of current epoch of the alpha file, let's call it the
770 find all files on disk
772 remove all files registered in the recentworld up to recent-ts
774 remove all files that have been deleted after recent-ts according to
777 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
779 * rersyncrecent might be a cronjob with a (locked) state file which
780 contains things like after and maybe last z sync or such?
782 rrr-mirror might be an alternative name but how would we justify the
783 three Rs when there is no Re-Rsync-Recent?
785 With the --loop parameter it is an endless loop, without it is no loop.
786 At least this is simple.
788 * todo: new accssor z-interval specifies how often the Z file is updated
789 against the filesystem. We probably want no epoch stamp on these
790 entries. And we want to be able to filter the entries (e.g. no
791 by-modules and by-category tree)
793 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
795 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
796 into Z. We do not need the other files filled up with filesystem
799 * need interface to query for a file in order to NOT call update on
800 PAUSE a second time within a short time.
802 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
804 * recommended update interval? Makes no sense, is different for
811 change-log-default-name: "Todo"