1 2013-08-04 04:10:55 <gmaxwell> FWIW, looks like TechLover is #bitcoin is another debian user getting a sync hang like reported in https://github.com/bitcoin/bitcoin/issues/2726
2 2013-08-04 04:11:30 <gmaxwell> er s/is/in/
3 2013-08-04 04:14:09 <coingenuity> gmaxwell: that's the thing i was bitching about for a while, though it hasn't returned since i started using 0.8.3 for some reason
4 2013-08-04 04:15:28 <coingenuity> i found via wiping out the block db and starting from a single connected peer over private link got it to the point that i could then index normally, strangely...
5 2013-08-04 04:16:18 <gmaxwell> some people have reported that setting par=1 is fixing it.
6 2013-08-04 04:16:40 <coingenuity> what's the par flag do? never used it before
7 2013-08-04 04:16:57 <gmaxwell> makes it use one thread for validation.
8 2013-08-04 04:17:02 <coingenuity> huh
9 2013-08-04 04:17:20 <coingenuity> thats.....strange that it's fixing the problem for some users
10 2013-08-04 04:17:30 <gmaxwell> Which suggests that the bug is somehow threading related. E.g. malloc arena being corrupted by multiple concurrent allocations.
11 2013-08-04 04:17:38 <warren> boost related?
12 2013-08-04 04:18:05 <coingenuity> i would have to guess that its somethign wrong with debian's impl of task scheduling
13 2013-08-04 04:18:17 <gmaxwell> warren: so, we statically link boost in our official binaries, I believe. And at least for edam on that pull it also fails with the official binaries.
14 2013-08-04 04:18:49 <gmaxwell> which suggests that the interaction is libc or the kernel; and doesn't really tell us if the bug is ours or something elses.
15 2013-08-04 04:19:05 <coingenuity> perhaps someone needs to audit debians task scheduler for crypto-friendliness
16 2013-08-04 04:19:13 <coingenuity> might be hash-dosing itself or something retarded
17 2013-08-04 04:19:40 <gmaxwell> IIRC data-race-detector had some complaints on the parallel validation stuff, but IIRC the complaints were in boost so it was hard to tell if it was just false positives.
18 2013-08-04 04:19:53 <Luke-Jr> gmaxwell: it's not a validation failure?
19 2013-08-04 04:20:08 <gmaxwell> coingenuity: nah, no amount of "dos" should permit this outcome. :P
20 2013-08-04 04:20:15 <Luke-Jr> or the validation threads getting stuck on some mutex?
21 2013-08-04 04:20:23 <gmaxwell> Luke-Jr: it just rejects a random block with a validation falure shortly after the top checkpoint.
22 2013-08-04 04:20:38 <coingenuity> http://www.anchor.com.au/blog/2012/12/how-to-explain-hash-dos-to-your-parents-by-using-cats/ is what i was talking about gmaxwell - just a phrase
23 2013-08-04 04:20:39 <Luke-Jr> gmaxwell: sounds like something isn't thread-safe in the validation process :x
24 2013-08-04 04:20:42 <gmaxwell> and seems to do so pretty reliably for the few users its doing it to.
25 2013-08-04 04:21:08 <gmaxwell> Luke-Jr: more complicated than that: never fails for most people, always fails for these people. All of these people are running recent debian.
26 2013-08-04 04:21:13 <gmaxwell> (AFAIK)
27 2013-08-04 04:21:25 <Luke-Jr> always fails the same way/block? O.o
28 2013-08-04 04:21:56 <coingenuity> Luke-Jr: thats the weirdness, it appears to depend on the particular client running the node
29 2013-08-04 04:22:01 <coingenuity> (from what i've seen)
30 2013-08-04 04:22:06 <Luke-Jr> recent = stable or testing? or unstable? :x
31 2013-08-04 04:22:26 <coingenuity> Luke-Jr: i've seen this in 0.8.x until .2, guess its still present in .3 as well
32 2013-08-04 04:22:53 <Luke-Jr> Debian packages? Official Ubuntu binaries? Or from source? or a mix?
33 2013-08-04 04:23:15 <coingenuity> sourceforge bins, user compiles, etc
34 2013-08-04 04:23:24 <coingenuity> just seems to be debian-flavored as the only real overtone
35 2013-08-04 04:23:25 <warren> .2 and .3 are almost the same thing
36 2013-08-04 04:25:12 <Luke-Jr> coingenuity: it happens every reindex?
37 2013-08-04 04:25:31 <warren> come to think of it, we've had similar reports from our users.
38 2013-08-04 04:25:33 <coingenuity> Luke-Jr: in my case, i couldn't get it to fully index all the way
39 2013-08-04 04:25:50 <coingenuity> Luke-Jr: i could only manage to get ~130k blocks before it'd start acting up
40 2013-08-04 04:25:56 <Luke-Jr> coingenuity: have you tried the first master revision with parallelized verification?
41 2013-08-04 04:26:04 <coingenuity> and then it'd reach a "bad block" and get stuck entirely
42 2013-08-04 04:26:21 <coingenuity> Luke-Jr: i just built a new dedi box for bitcoin and the problem has vanished :/
43 2013-08-04 04:26:33 <Luke-Jr> coingenuity: argh, so you can't reproduce anymore? :\\
44 2013-08-04 04:26:51 <coingenuity> i totally decomissioned the old box due to the lack of reliability from bitcoind
45 2013-08-04 04:27:05 <coingenuity> just built new hardware and restarted
46 2013-08-04 04:27:36 <Luke-Jr> jerk <.<
47 2013-08-04 04:27:38 <Luke-Jr> :p
48 2013-08-04 04:27:52 <coingenuity> anyway, once i got it to 95% full of blocks with a local seed peer it stopped with the misbehavior on index
49 2013-08-04 04:28:26 <coingenuity> but there was still block-related crashing even during synced operation which would trigger a full reindex...so i just scrapped the machine
50 2013-08-04 04:28:45 <coingenuity> (yes, it was almost-new hardware)
51 2013-08-04 05:58:44 <Luke-Jr> coingenuity: btw, did you confirm the problem was still there after a fresh reinstall?
52 2013-08-04 07:18:32 <coingenuity> Luke-Jr: yeah, i wiped out everything except wallet.dat and the problem still occured
53 2013-08-04 07:18:34 <coingenuity> however, problem occurred at much lower rates when using a different wallet
54 2013-08-04 07:18:35 <coingenuity> s/different/empty/
55 2013-08-04 07:27:56 <sipa> have you ran memtest86 on that machine?
56 2013-08-04 08:18:01 <coingenuity> sipa: considering it is(was) brand new hardware i didnt bother debugging too heavily on that side
57 2013-08-04 08:19:33 <warren> coingenuity: even new hardware can be bad
58 2013-08-04 08:19:39 <warren> memtest86 is easy
59 2013-08-04 14:06:01 <PK> I have a feature request. Often when I start bitcoind and catch up with the block chain, it gets corrupted and I have reindex the whole thing. Can you add the option to specify from which block the reindexing shall start?
60 2013-08-04 14:06:50 <MC1984_> that shouldnt happen
61 2013-08-04 14:07:04 <PK> for something that shouldn't happen, it happens a lot to me though :(
62 2013-08-04 14:07:15 <QuantumQrack> never happens to me
63 2013-08-04 14:07:21 <MC1984_> post debug.log
64 2013-08-04 14:10:09 <PK> this is the interesting part (probably) with the last block it accepted: http://pastebin.com/79gVurLA
65 2013-08-04 14:13:03 <MC1984_> doesnt look interesting
66 2013-08-04 14:13:10 <MC1984_> nothing about database problems
67 2013-08-04 14:18:00 <_ingsoc> Where does one go to get a crypto algorithm developed these days?
68 2013-08-04 14:45:19 <jgarzik> who develops new algorithms? :)
69 2013-08-04 14:45:23 <jgarzik> that takes years
70 2013-08-04 14:49:19 <sipa> PK: you cannot start reindex anywhere from the beginning
71 2013-08-04 14:49:34 <sipa> PK: the point is that there database is corrupted, so you have to throw it away and rebuild it from scratch
72 2013-08-04 14:50:18 <phantomcircuit> PK, that says nothing about any real errors
73 2013-08-04 14:50:29 <phantomcircuit> it's just preventing transactions from being added to the memory pool
74 2013-08-04 14:50:34 <phantomcircuit> which is 100% irrelevant
75 2013-08-04 14:51:59 <sipa> PK: what hardware and OS and bitcoin version?
76 2013-08-04 15:50:39 <Vinnie_win> wuddup folks
77 2013-08-04 16:42:58 <ahmedbodi> can someone here recommend a sha based stratum i can use?
78 2013-08-04 19:54:03 <PK> sipa: win xp 64, hdd (not ssd), core2 quad cpu.
79 2013-08-04 19:58:13 <PK> I give it another try tomorrow
80 2013-08-04 20:10:39 <nsh> jgarzik, do you know anyone that might have time to do a some reveng on that FBI js->FF exploit?
81 2013-08-04 20:11:40 <jgarzik> nsh, no. Security people aren't really in my circles for no particular reason
82 2013-08-04 20:11:56 <nsh> so far it seems to be doing a few heap sprays and using ws2_32.dll & iphlapi.dll to make a clearnet http request with the UUID token
83 2013-08-04 20:12:00 <nsh> ok, ty anyway
84 2013-08-04 20:12:09 <nsh> it'll come out in the wash, just curious :)
85 2013-08-04 20:13:14 <nsh> everyone's a bit defcon'd still i think
86 2013-08-04 20:13:19 <nsh> or ohm'd
87 2013-08-04 20:20:29 <ahmedbodi> does anyone here know a good stratum which is compatible with bitcoin
88 2013-08-04 20:21:04 <Luke-Jr> ahmedbodi: please rephrase your question
89 2013-08-04 20:21:39 <sipa> i assume he's looking for a mining pool that supports stratum
90 2013-08-04 20:22:07 <ahmedbodi> basically im after a copy of stratum which will work with all sha256 based coins for use with my pools, i normally use moopless's stratum-mining-litecoin but thats incompatible with sha based coins
91 2013-08-04 20:22:33 <sipa> stratum is a protocol
92 2013-08-04 20:22:44 <sipa> a copy of stratum doesn't mean muvh
93 2013-08-04 20:22:47 <sipa> much
94 2013-08-04 20:23:01 <Luke-Jr> ahmedbodi: You want a single poolserver that serves multiple coins at once?
95 2013-08-04 20:23:23 <Luke-Jr> I don't think that's even a goal for anyone; and not possible with a single stratum port
96 2013-08-04 20:23:39 <ahmedbodi> nooo
97 2013-08-04 20:24:00 <ahmedbodi> basically https://github.com/moopless/stratum-mining-litecoin
98 2013-08-04 20:24:08 <ahmedbodi> something like that but for sha based coins
99 2013-08-04 20:25:00 <sipa> https://github.com/slush0/stratum-mining
100 2013-08-04 20:25:01 <Luke-Jr> ahmedbodi: Eloipool?
101 2013-08-04 20:25:19 <ahmedbodi> i did look at your implementation of eloipool Luke-Jr, only problem is compatibilty with my frontend i think
102 2013-08-04 20:25:35 <ahmedbodi> does it work with standard mmcfe?
103 2013-08-04 20:26:16 <Luke-Jr> ahmedbodi: no idea; the question is does mmcfe work with "standard" Eloipool :p
104 2013-08-04 20:26:38 <ahmedbodi> that question will do
105 2013-08-04 20:26:54 <ahmedbodi> does mmcfe work with the standard eloipool?
106 2013-08-04 20:27:22 <ahmedbodi> we can blame the speech barriers on american english haha
107 2013-08-04 20:27:26 <Luke-Jr> I don't even know what mmcfe is
108 2013-08-04 20:27:42 <ahmedbodi> its a frontend for mining pools
109 2013-08-04 20:27:54 <Luke-Jr> I would guess it's a matter of configuration
110 2013-08-04 20:27:56 <ahmedbodi> http://www.hobonickle.com/
111 2013-08-04 20:28:05 <ahmedbodi> like that, which is my current pool :D
112 2013-08-04 20:28:20 <sipa> ahmedbodi: i have no idea what that stratum-mining does, but the github project was forked off that project by slush for which i gave you the link
113 2013-08-04 20:28:36 <sipa> my guess is that it is very similar, but for bitcoin
114 2013-08-04 20:28:52 <ahmedbodi> yeah the one i gave is a litecoin version which has been cleaned up and optimized
115 2013-08-04 20:29:58 <Luke-Jr> sipa: I don't think slush's version does any kind of pool logging
116 2013-08-04 20:30:16 <ahmedbodi> my main problem with eloipool is if it supports pushpool's db structure that way it wont take me long to modify the db to fit my frontend
117 2013-08-04 20:33:32 <ahmedbodi> Luke-Jr: ^
118 2013-08-04 20:34:04 <Luke-Jr> ahmedbodi: Eloipool is fully configurable for the database. Just set it up to use pushpool's schema.
119 2013-08-04 20:34:23 <ahmedbodi> i see,
120 2013-08-04 20:34:31 <ahmedbodi> im guessing i have to set that in the config file
121 2013-08-04 20:44:03 <ahmedbodi> heres what the schema is for pushpool
122 2013-08-04 20:44:20 <ahmedbodi> "stmt.pwdb":"SELECT password FROM pool_worker WHERE username = ?",
123 2013-08-04 20:44:21 <ahmedbodi> "stmt.sharelog":"INSERT INTO shares (rem_host, username, our_result, upstream_result, reason, solution) VALUES (?, ?, ?, ?, ?, ?)"
124 2013-08-04 20:44:23 <ahmedbodi> },
125 2013-08-04 20:45:30 <sipa> perhaps #bitcoin-mining ?
126 2013-08-04 20:45:47 <ahmedbodi> i see, i didnt know that existed, thanks
127 2013-08-04 20:47:47 <ahmedbodi> as far as i can tell Luke-Jr's eloipool will work with the database perfectly, just one small little error for resolving
128 2013-08-04 20:49:09 <Luke-Jr> ahmedbodi: there is an example config for Eloipool
129 2013-08-04 20:49:19 <ahmedbodi> yeah thats what ive done a comparison to
130 2013-08-04 20:49:27 <ahmedbodi> and its almost working apart from the db part
131 2013-08-04 20:49:59 <ahmedbodi> the fronend reads the our_result and upstream_result as 0/1 wheres eloipool puts them as Y/N
132 2013-08-04 20:50:26 <Luke-Jr> that's odd, pushpool always used Y/N
133 2013-08-04 20:51:38 <ahmedbodi> give me a min to check my current db
134 2013-08-04 20:52:41 <ahmedbodi> never mind i seem to be misinformed, sorry about that
135 2013-08-04 20:52:52 <ahmedbodi> now all i need to get working is the user auth somehow
136 2013-08-04 20:58:09 <ahmedbodi> any ideas Luke-Jr
137 2013-08-04 20:59:21 <Luke-Jr> ahmedbodi: kinlo would know
138 2013-08-04 21:00:05 <ahmedbodi> i see, thanks bro, i gotta say, youve done some good work on eloipool from what ive seen