2010-06-23 04:27:24 utc |
mnichols |
Hi all |
2010-06-23 04:29:40 utc |
mnichols |
Question for the collective... Is it normal to have 180mb in mu msgs dir? |
2010-06-23 04:30:43 utc |
mnichols |
I thought that was supposed to be kind of transient. ?? |
2010-06-23 04:33:21 utc |
mnichols |
I also have several message files that say something like "-bash: !23456248307460: event not found". Not sure what that means.?? |
2010-06-23 05:07:15 utc |
jmettraux |
hello |
2010-06-23 05:07:24 utc |
jmettraux |
msgs should be empty |
2010-06-23 05:07:45 utc |
jmettraux |
it's supposed to be consumed on the fly |
2010-06-23 05:08:05 utc |
jmettraux |
if you're sure that your engine is OK, you can wipe out that msgs dir |
2010-06-23 05:08:35 utc |
jmettraux |
for the "-bash !xxx" thing, do you have more details ? |
2010-06-23 05:31:26 utc |
mnichols |
Sorry. Didn't see your reply. |
2010-06-23 05:32:05 utc |
mnichols |
I just rebuilt my server and started up a worker and it looks like it is consuming the msgs. |
2010-06-23 05:32:33 utc |
jmettraux |
great |
2010-06-23 05:32:53 utc |
mnichols |
Those msg files with the -bash message... That's all it says. Is there something else I can look at? |
2010-06-23 05:34:44 utc |
jmettraux |
could you please gist one of those ? |
2010-06-23 05:35:08 utc |
mnichols |
k |
2010-06-23 05:36:27 utc |
mnichols |
http://gist.github.com/449552 |
2010-06-23 05:36:49 utc |
jmettraux |
woah |
2010-06-23 05:36:57 utc |
mnichols |
I have 322 similar messages. |
2010-06-23 05:37:01 utc |
jmettraux |
never seen such a symptom |
2010-06-23 05:37:05 utc |
mnichols |
(that I've found) |
2010-06-23 05:37:56 utc |
jmettraux |
ruote doesn't have any "event not found" string |
2010-06-23 05:38:10 utc |
mnichols |
Spooky |
2010-06-23 05:38:57 utc |
mnichols |
Might if have something to do with the cloche file locks? Maybe out of file descriptors??? |
2010-06-23 05:39:11 utc |
mnichols |
(stab in the dark) |
2010-06-23 05:39:14 utc |
jmettraux |
maybe |
2010-06-23 05:45:23 utc |
mnichols |
Wow... worker is really flying through these messages now. I think I may have had some systemic problems with old server. The "-bash" leads me to believe there was something wrong at the resource/system level. |
2010-06-23 05:45:59 utc |
jmettraux |
how many processes do you usually have ? |
2010-06-23 05:47:04 utc |
mnichols |
We go in large bursts. Right now I'm around 100k |
2010-06-23 05:47:12 utc |
mnichols |
Up to 200k |
2010-06-23 05:48:33 utc |
jmettraux |
wow |
2010-06-23 05:49:07 utc |
mnichols |
I have it on an EBS Volume. I could share a snapshot with you if it would be helpful to you. |
2010-06-23 05:49:36 utc |
jmettraux |
I'm afraid I wouldn't have to time to look at it :-( |
2010-06-23 05:50:02 utc |
mnichols |
np |
2010-06-23 05:50:32 utc |
jmettraux |
have the time |
2010-06-23 05:51:18 utc |
mnichols |
I'll keep an eye out for the issue again. Anything in particular you think I should monitor? |
2010-06-23 05:52:11 utc |
jmettraux |
I would need more context |
2010-06-23 05:52:30 utc |
jmettraux |
if you could give me instructions to reproduce the issue |
2010-06-23 05:52:50 utc |
jmettraux |
on how to reproduce the issue |
2010-06-23 05:53:21 utc |
jmettraux |
nobody else reported such an issue |
2010-06-23 05:54:01 utc |
mnichols |
I'm clueless right now. Just trying to get through this big job. |
2010-06-23 05:54:30 utc |
mnichols |
Where do message errors normally appear? |
2010-06-23 05:54:56 utc |
jmettraux |
next to /msgs/, under /errors/ |
2010-06-23 05:55:04 utc |
mnichols |
Seems like there used to be a directory that they went in, but it went away with 2.0 |
2010-06-23 05:55:08 utc |
mnichols |
oh. |
2010-06-23 05:55:34 utc |
mnichols |
I don't have an errors dir |
2010-06-23 05:55:50 utc |
jmettraux |
do you suspect errors ? |
2010-06-23 05:56:41 utc |
mnichols |
No. I've tested my wf pretty well. But it seems that this would have triggered and "error". |
2010-06-23 05:57:25 utc |
mnichols |
Besides definition problems, what other types of errors could show up? |
2010-06-23 05:58:02 utc |
jmettraux |
runtime errors |
2010-06-23 05:58:08 utc |
jmettraux |
participants gone missing |
2010-06-23 05:58:13 utc |
jmettraux |
disk full |
2010-06-23 05:58:16 utc |
jmettraux |
... |
2010-06-23 05:58:21 utc |
mnichols |
ah |
2010-06-23 05:59:13 utc |
mnichols |
So if I were to set up a monitor, I should look at /msgs/errors for new files? |
2010-06-23 05:59:56 utc |
jmettraux |
at /errors |
2010-06-23 06:00:10 utc |
mnichols |
k |
2010-06-23 06:00:36 utc |
jmettraux |
or, in ruby, you could directly ask the engine |
2010-06-23 06:00:49 utc |
mnichols |
engine.errors ? |
2010-06-23 06:00:57 utc |
jmettraux |
p "ARGH" if engine.errors.size > 0 |
2010-06-23 06:01:01 utc |
jmettraux |
:) |
2010-06-23 06:01:08 utc |
mnichols |
) |
2010-06-23 06:01:11 utc |
mnichols |
:) |
2010-06-23 06:01:16 utc |
mnichols |
Got it. |
2010-06-23 06:01:28 utc |
mnichols |
Different subject... |
2010-06-23 06:01:50 utc |
mnichols |
I tried that storage engine migration utility that you added to .10 |
2010-06-23 06:02:05 utc |
mnichols |
It worked awesome... until I ran our of memory. :) |
2010-06-23 06:02:45 utc |
mnichols |
Looks like get_many loads all of the items. It choked on expressions. |
2010-06-23 06:03:04 utc |
jmettraux |
aaah |
2010-06-23 06:03:06 utc |
jmettraux |
too bad |
2010-06-23 06:03:22 utc |
jmettraux |
if you could document that failure in the mailing list |
2010-06-23 06:03:25 utc |
mnichols |
Thought I could just split up the dirs and take a chunk at a time. |
2010-06-23 06:03:59 utc |
jmettraux |
that would make up for a good TODO |
2010-06-23 06:04:32 utc |
mnichols |
I looked into a way to make it load chunks at a time, but that would require mods to all of the storage engines. They all take their own opts={} |
2010-06-23 06:05:55 utc |
mnichols |
Not a high priority for me right now. I was hoping to get my expressions into couch so I could do some fancy querying. |
2010-06-23 06:07:09 utc |
jmettraux |
querying on workitems or expressions ? |
2010-06-23 06:09:32 utc |
mnichols |
I guess workitems |
2010-06-23 06:09:42 utc |
mnichols |
Want to see current status |
2010-06-23 06:12:02 utc |
mnichols |
I've tried querying the engine, but FsStorage takes a while to pull up a process when there are a lot. |
2010-06-23 06:12:27 utc |
jmettraux |
I should do something about that |
2010-06-23 06:12:51 utc |
mnichols |
Wouldn't couch be faster? |
2010-06-23 06:13:09 utc |
mnichols |
At least for lookup? |
2010-06-23 06:13:58 utc |
jmettraux |
probably |
2010-06-23 06:14:19 utc |
jmettraux |
but I should find a way to speed up FsStorage anyway |
2010-06-23 06:15:37 utc |
mnichols |
I saw your blog post saying you are using ruote-couch in a couple of projects. What do you like about it? |
2010-06-23 06:16:19 utc |
jmettraux |
I use couch for the rest of the data, so everything goes in couch |
2010-06-23 06:16:24 utc |
jmettraux |
but it's slow... |
2010-06-23 06:16:34 utc |
mnichols |
oh |
2010-06-23 06:16:57 utc |
jmettraux |
http://ruote.rubyforge.org/configuration.html#storage |
2010-06-23 06:17:02 utc |
jmettraux |
maybe lookup is fast |
2010-06-23 06:17:26 utc |
jmettraux |
but the "it goes over the network and back" "via http" is present |
2010-06-23 06:17:34 utc |
mnichols |
right |
2010-06-23 06:17:56 utc |
mnichols |
I saw this. Redis is an interesting option. |
2010-06-23 06:18:44 utc |
mnichols |
I'm not sure what size constraints it has though. (have it on my todo list to check into it) |
2010-06-23 06:20:53 utc |
jmettraux |
redis < 2.0 requires you to have everything in memory |
2010-06-23 06:21:07 utc |
jmettraux |
redis >= 2.0 requires you to have all the keys in memory |
2010-06-23 06:21:14 utc |
mnichols |
I suspected. |
2010-06-23 06:21:36 utc |
jmettraux |
maybe the last message in there : http://groups.google.com/group/openwferu-users/browse_frm/thread/5cb80730419a9a38 might interest you |
2010-06-23 06:22:29 utc |
mnichols |
Oh, yeah I saw this. One of the reasons I was investigating couch. |
2010-06-23 06:23:39 utc |
mnichols |
What is stored in "workitems"? |
2010-06-23 06:24:08 utc |
jmettraux |
workitems |
2010-06-23 06:24:16 utc |
mnichols |
:) |
2010-06-23 06:24:18 utc |
jmettraux |
well |
2010-06-23 06:24:21 utc |
mnichols |
Really? |
2010-06-23 06:24:35 utc |
jmettraux |
the once managed by Ruote::StorageParticipant participants |
2010-06-23 06:25:39 utc |
jmettraux |
the ones... |
2010-06-23 06:25:51 utc |
mnichols |
So why do I not have a workitems dir? |
2010-06-23 06:25:52 utc |
jmettraux |
(sorry writing in another language in other windows) |
2010-06-23 06:25:57 utc |
mnichols |
:) |
2010-06-23 06:25:58 utc |
mnichols |
np |
2010-06-23 06:26:10 utc |
jmettraux |
maybe you don't use a Ruote::StorageParticipant |
2010-06-23 06:26:48 utc |
mnichols |
Not required? |
2010-06-23 06:29:22 utc |
mnichols |
Where would workitems be stored if Ruote::StorageParticipant is not used? |
2010-06-23 06:30:00 utc |
jmettraux |
if you didn't register any Ruote::StorageParticipant no need for /workitems |
2010-06-23 06:30:15 utc |
jmettraux |
I guess you have participants that use them directly and so and do |
2010-06-23 06:30:17 utc |
jmettraux |
and so |
2010-06-23 07:07:54 utc |
jmettraux |
tosch_le: hello Torsten |
2010-06-23 07:08:06 utc |
tosch_le |
hello john! |
2010-06-23 07:08:41 utc |
jmettraux |
how are you doing ? |
2010-06-23 07:08:57 utc |
tosch_le |
fine, thanks. and you? |
2010-06-23 07:09:18 utc |
jmettraux |
doing fine :) |
2010-06-23 07:14:59 utc |
mnichols |
Thanks for you help, John. Talk to you later. |
2010-06-23 07:15:14 utc |
jmettraux |
mnichols: you're welcome ! |
2010-06-23 07:42:41 utc |
lbt |
o/ |
2010-06-23 16:23:15 utc |
lbt |
woohoo... a python participant :) |
2010-06-23 16:36:07 utc |
kennethkalmer |
lbt: awesome stuff ! |
2010-06-23 16:36:25 utc |
kennethkalmer |
just merge jmettraux & your work into my branch |
2010-06-23 16:36:28 utc |
kennethkalmer |
and push to github |
2010-06-23 16:36:38 utc |
kennethkalmer |
can I push a new ruote-amqp gem to gemcutter ? |
2010-06-23 17:18:21 utc |
lbt |
kennethkalmer: I think we need to check with john before cutting another gem |
2010-06-23 17:18:40 utc |
lbt |
I'd like to put the python code into r-a though |
2010-06-23 17:19:49 utc |
kennethkalmer |
np on checking with john |
2010-06-23 17:20:00 utc |
kennethkalmer |
i'm very skeptical on having python code in the repo |
2010-06-23 17:20:12 utc |
lbt |
why? :) |
2010-06-23 17:20:22 utc |
lbt |
it's ruote... not ruby :) |
2010-06-23 17:20:25 utc |
lbt |
? |
2010-06-23 17:20:34 utc |
kennethkalmer |
it is ruby :) |
2010-06-23 17:20:38 utc |
lbt |
heh |
2010-06-23 17:20:51 utc |
kennethkalmer |
why not publish it as a separate repo/egg ? |
2010-06-23 17:21:02 utc |
kennethkalmer |
ACTION python-fu is very low |
2010-06-23 17:21:05 utc |
lbt |
well, we could... |
2010-06-23 17:21:19 utc |
kennethkalmer |
i think it would be best |
2010-06-23 17:21:23 utc |
lbt |
here: http://gist.github.com/450248 |
2010-06-23 17:21:40 utc |
lbt |
it's almost ruby |
2010-06-23 17:22:03 utc |
lbt |
except the comments come after the def and not before... |
2010-06-23 17:22:31 utc |
kennethkalmer |
:) |
2010-06-23 17:22:33 utc |
lbt |
s/end// <---- lbt's patented ruby2python converter |
2010-06-23 17:22:46 utc |
kennethkalmer |
i think a separate project is best |
2010-06-23 17:22:48 utc |
lbt |
OK |
2010-06-23 17:22:58 utc |
kennethkalmer |
thanks for understanding |
2010-06-23 17:23:06 utc |
kennethkalmer |
pretty cool though |
2010-06-23 17:23:10 utc |
kennethkalmer |
have to admit it kicks ass |
2010-06-23 17:24:30 utc |
lbt |
bbiab ... chicken+mango kebabs.... mmm |
2010-06-23 17:25:23 utc |
kennethkalmer |
enjoy |
2010-06-23 17:25:29 utc |
kennethkalmer |
i'm also starting dinner in a bit |
2010-06-23 18:41:45 utc |
lbt |
kennethkalmer: one thing I did find... r-a is using the anonymous exchange; ie "" |
2010-06-23 18:42:35 utc |
lbt |
thinking it should probably use "ruote-participants" or something |