1 2008-08-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
3 * deletes: if a delete follows an add quickly enough it may happen that
4 a downstream mirror did not see the add at all! It seems this needs to
5 be mentioned somewhere.
7 2008-08-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
9 * I suspect the treat of metadata is incorrect during read or something.
10 The bug that I am watching is that between 06:08 and 06:09 the 6h file
11 contained more than 6 hours worth of data. At 06:08 we merged into the
12 1d file. We need to take snapshots of the 6h file over the course of an
13 hour or maybe only between XX:08 and XX:09? Nope, the latter is not
16 Much worse: watching the 1h file: right at the moment (at 06:35) it
17 covers 1218867584-1219120397 which is 70 hours.
19 Something terribly broken. BTW, 1218867584 corresponds to Sat Aug 16
20 08:19:44 2008, that is when I checked out last time, so it seems to be
21 aggregating and never truncating?
23 No, correct is: it is never truncating; but wrong is: it is aggregating.
24 It does receive a lot of events from time to time from a larger file.
25 Somehow a large file gets merged into the small one and because the
26 "meta/merged" attribute is missing, nobody is paying attention. I
27 believe that I can fix this by making sure that metadata are honoured
28 during read. DONE and test adjusted.
30 2008-08-17 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
34 remotebase => remoteroot to fit well with localroot DONE
35 local_path() => localroot seems to me should already work DONE
36 recentfile_basename => rfilename no need to stress it has no slash DONE
38 filenameroot??? Doesn't seem too bad to me today. Maybe something like
39 kern? It would anyway need a deprecation cycle because it is an
40 important constructor.
42 * I like the portability that Data::Serializer brings us but the price
43 is that some day we might find out that it is slowing us a bit. We'll
46 2008-08-16 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
48 * should we not enter the interval of the principal (or the interval of
49 the merging file?) in every aggregated/merged file?
51 * we should aim at a first release and give up on thinking about
52 sanitizing stuff and zloop. Let's just admit that a full traditional
53 rsync is the only available sanitizer ATM. Otherwise it's complicated
54 stuff: sanitizing on the origin server, sanitizing on the slaves,
55 sanitizing forgotten files, broken timestamps, etc. Let's delay it and
56 get the basics out before this becomes a major cause for mess.
58 2008-08-13 Andreas Koenig <k@andreas-koenigs-computer.local>
60 * On OSes not supporting symlinks we expect that RECENT.recent contains
61 the contents of the principal recentfile. Actually this is identical on
62 systems supporting symlinks. Simple, what follows from that is that we
63 need to keep the serializer in the metadata because we cannot read it
64 from the filename, doesn't it? Of course not. It's a chicken and egg
65 problem. This leaves us with the problem to actually parse the
66 serialized data to find out in which format it is. So who can do the 4
67 or 5 magics we wanted to support? File::LibMagic?
69 2008-08-09 Andreas Koenig <k@andreas-koenigs-computer.local>
71 * remotebase and recentfile_basename are ugly names. Now that we need a
72 word for the shortest/principal/driving recentfile too we should do
75 localroot is good. rfile is good. local_path() is bad, local_path($path)
76 is medium, filenameroot() is bad, remotebase is bad, recentfile is
79 Up to now remotebase was the string that described the remote root
80 directory in rsync notation, like pause.perl.org::authors. And
81 recentfile_basename was "RECENT-1h.yaml".
83 2008-08-08 Andreas Koenig <k@andreas-koenigs-computer.local>
85 * The test that was added in today's checkin is a good start for a test
86 of rmirror. We should have more methods in Recent.pm: verify,
87 addmissingfiles. We should verify the current tree, then rmirror it and
88 then verifytree the copy. We could then add some arbitrary file and let
89 it be discovered by addmissingfiles, then rmirror again and then
90 verifytree the copy again.
92 Then we could start stealing from csync2 sqlite database [no port to
93 OSX!] and fill a local DB. And methods to compare the database with the
94 recentfiles. Our strength is that in principle we could maintain state
95 with a single float. We have synced up to 1234567890.123456. If the Z
96 file does not add new files all we have to do is mirror the new ones and
99 This makes it clear that we should extend current protocol and declare
100 that we cheat when we add files too late, just to help the other end
101 keeping track. Ah yes, that's what was meant when zloop was mentioned
104 Maybe need to revisit File::Mirror to help me with this task.
106 2008-08-07 Andreas Koenig <k@andreas-koenigs-computer.local>
108 * There must be an allow-me-to-truncate flag in every recentfile.
109 Without it one could construct a sequence of updates winning the locking
110 battle against the aggregator. Only if an aggregator has managed to
111 merge data over to the next level, truncating can be allowed. DONE with
114 2008-08-06 Andreas Koenig <k@andreas-koenigs-computer.local>
116 * We should probably guarantee that no duplicates enter the aggregator
119 2008-08-02 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
121 * To get merge operation faster would need a good benchmark test. What
122 02 spits out isn't reliable enough and is dominated by many other
125 commit 10176bf6b79865d4fe9f46e3857a3b8669fa7961
126 Author: Andreas J. Koenig <k@k75.(none)>
127 Date: Sat Aug 2 07:58:04 2008 +0200
131 commit 3243120a0c120aaddcd9b1f4db6689ff12ed2523
132 Author: Andreas J. Koenig <k@k75.(none)>
133 Date: Sat Aug 2 11:40:29 2008 +0200
135 there was a lot of trying but the effect is hardly measurable with
138 * overhead of connecting seems high. When setting
139 max_files_per_connection to 1 we see that.
141 2008-08-01 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
143 * 1217622571.0889 - 1217597432.86734 = 25138.2215600014
145 25138.2215600014/3600 = 6.98283932222261
147 It jumps into the eye that this is ~ 7 hours, not ~6, so there seems to
148 be a bug in the aggregator. FIXED
150 2008-07-27 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
152 * e.g. id/Y/YE/YEWENBIN/Emacs-PDE-0.2.16.tar.gz: Do we have it, should
153 we have it, can we mirror it, mirror it!
155 I fear this needs a new class which might be called
156 File::Rsync::Mirror::Recent. It would collect all recentfiles of a kind
157 and treat them as an entity. I realize that a single recentfile may be
158 sufficient for certain tasks and that it is handy for the low level
159 programmer but it is not nice to use. If there is a delete in the 1h
160 file then the 6h file still contains it. Seekers of the best information
161 need to combine at least some of the recentfiles most of the time.
163 There is the place for the Z loop!
165 But the combination is something to collect in a database, isn't it. Did
166 csync2 just harrumph?
168 2008-07-26 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
170 * it just occurred to me that hosts in the same mirroring pool could
171 help out each other even without rewriting the recentfile. Just fetch
172 the stuff to mirror from several places, bingo. But that's something
173 that should rather live in a separate package or in rsync directly.
175 * cronjobs are unsuited because with ntp they would all come at the full
176 minute and disturb each other. Besides that I'd hate to have a backbone
177 with more than a few seconds latency.
179 2008-07-25 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
181 * a second rsync server with access control for PAUSE. Port? 873 is the
182 standard port, let's take 8873.
184 * if there were a filesystem based on this, it would have a slow access
185 to inexistent files. It would probably provide wrong readdir (only based
186 on current content) or also a slow one (based on a recentfile written
187 after the call). But it would provide fast access to existing files. Or
188 one would deliberately allow slightly blurred answers based on some
189 sqlite reflection of the recentfiles.
191 * todo: write a variant of mirror() that combines two or more
192 recentfiles and treats them like one
194 * todo: signal handler to remove the tempfile
196 2008-07-24 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
198 * now that we have the symlink I forgot how it should be used in
201 * the z loop: add missing files to Z file. Just append them (instead of
202 prepending). So one guy prepends something from the Y file from time to
203 time and another guy appends something rather frequently. Collecting
204 pond. When Y merges into Z, things get epoch and the collecting pond
205 gets smaller. What exactly are "missing files"?
207 take note of current epoch of the alpha file, let's call it the
210 find all files on disk
212 remove all files registered in the recentworld up to recent-ts
214 remove all files that have been deleted after recent-ts according to
217 2008-07-23 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
219 * rersyncrecent might be a cronjob with a (locked) state file which
220 contains things like after and maybe last z sync or such?
222 rrr-mirror might be an alternative name but how would we justify the
223 three Rs when there is no Re-Rsync-Recent?
225 With the --loop parameter it is an endless loop, without it is no loop.
226 At least this is simple.
228 * todo: new accssor z-interval specifies how often the Z file is updated
229 against the filesystem. We probably want no epoch stamp on these
230 entries. And we want to be able to filter the entries (e.g. no
231 by-modules and by-category tree)
233 2008-07-20 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
235 * Fill the Z file. gc or fsck or both. Somehow we must get the old files
236 into Z. We do not need the other files filled up with filesystem
239 * need interface to query for a file in order to NOT call update on
240 PAUSE a second time within a short time.
242 2008-07-19 Andreas J. Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de>
244 * recommended update interval? Makes no sense, is different for
251 change-log-default-name: "Todo"