Created
July 13, 2012 08:19
-
-
Save galderz/3103593 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Jul 09 17:13:59 <galderz> ping vblagoje | |
Jul 09 17:14:07 <vblagoje> hey galderz | |
Jul 09 17:14:21 <galderz> vblagoje, i forgot to ask earlier about https://github.com/infinispan/infinispan/pull/1185 | |
Jul 09 17:14:34 <galderz> you said you'll provide a new pull req | |
Jul 09 17:15:00 <galderz> where are you with this? i guess it's lower prio than the M/R+Cache Loader stuff? | |
Jul 09 17:15:04 <vblagoje> yes, it is still a lot of work and in new pull I'll reference this one | |
Jul 09 17:15:16 <galderz> vblagoje, ok - was just wondering if you were finished with it :) | |
Jul 09 17:15:20 <vblagoje> yeah, looking into cache loader now | |
Jul 09 17:15:41 <vblagoje> yeah, it is more or less finished, needs polishing; but after cache loader stuff is done | |
Jul 09 17:16:22 <galderz> vblagoje, what about https://github.com/infinispan/infinispan/pull/1207 ? | |
Jul 09 17:16:36 <galderz> vblagoje, are you waiting for sandros' feedback first? | |
Jul 09 17:17:02 <vblagoje> that should be ok; yes realistically should but he is on vacation for another 7 days or so | |
Jul 09 17:17:16 <galderz> vblagoje, ok, i'll integrate it then | |
Jul 09 17:17:32 <vblagoje> though about putting this into next alpha and having him reopen if this does not work - it really should | |
Jul 09 17:18:49 <galderz> vblagoje, did you consider adding a test for the issue he had? | |
Jul 09 17:19:00 <galderz> vblagoje, that'd make sure it doesn't fall through the cracks again | |
Jul 09 17:19:34 <vblagoje> it is hard; as we need AS to be there I guess | |
Jul 09 17:20:12 <vblagoje> he had this issue only in integration with Infinispan and AS and invoking m/r there | |
Jul 09 17:20:40 <galderz> vblagoje, how come? sounds a bit weird looking at the stacktrace | |
Jul 09 17:20:52 <galderz> sounds like the issue might be cos the cache was configured with lazy deserialization on? | |
Jul 09 17:20:55 <galderz> java.lang.ClassCastException: org.infinispan.marshall.MarshalledValue cannot be cast to de.adorsys.fireonec2.server.PrimeStats | |
Jul 09 17:21:20 <vblagoje> yeah, not sure tbh galderz | |
Jul 09 17:21:32 <galderz> vblagoje, welll, if you're not sure, how can you have a fix for it? :) | |
Jul 09 17:21:34 <vblagoje> i do not know how he ended up with those marshalled values | |
Jul 09 17:21:59 <vblagoje> well, his fix made invocation context local | |
Jul 09 17:22:06 <galderz> vblagoje, those marshalled values appear if the cache is configured with storeAsBinary | |
Jul 09 17:22:07 <galderz> :) | |
Jul 09 17:22:15 <vblagoje> my fix is just a nicer version of his | |
Jul 09 17:22:27 <galderz> vblagoje, that doesn't mean that his fix is right | |
Jul 09 17:22:52 <galderz> vblagoje, the issue I have with this is that the problem is unknown and wo/ it, i don't see how we can fix it | |
Jul 09 17:23:12 <vblagoje> true, however, after I have thought about this invocation context, it should never be constructed remote but always as local | |
Jul 09 17:23:21 <galderz> vblagoje, so, something simple you can try to do is to see if you can send a map/reduce function to a cache configured with storeAsBinary | |
Jul 09 17:24:11 <galderz> vblagoje, what made you think that? | |
Jul 09 17:24:58 <galderz> vblagoje, i think we can further improve the quality of the code if: | |
Jul 09 17:25:23 <vblagoje> as m/r invocation comes to remote vm it should always load data from local context; no need for remote | |
Jul 09 17:25:24 <galderz> you revert the fix, you add a test to send a map/reduce function to a cache configured with storeAsBinary=true, and see if you can replucate it | |
Jul 09 17:25:46 <vblagoje> sure | |
Jul 09 17:25:46 <galderz> if you can, then you've found your test, and you can then guarantee that the fix does its job :) | |
Jul 09 17:25:56 <vblagoje> makes sense | |
Jul 09 17:28:55 <galderz> vblagoje, thx :) | |
Jul 09 17:29:17 <vblagoje> np, galderz thx for suggestion | |
Jul 09 17:30:34 * mmarkus (~mmarkus@redhat/jboss/mmarkus) has left #infinispan | |
Jul 09 17:32:52 <vblagoje> galderz, do you know if AS configures caches with storeAsBinary by default | |
Jul 09 17:34:49 <galderz> vblagoje, some of them yeah, i.e. http session repl, 2LC entity caches... | |
Jul 09 17:35:03 <galderz> i've no idea which cache this guy uses | |
Jul 09 17:35:20 <galderz> vblagoje, it would have been handy for him to upload the example that was failing for him | |
Jul 09 17:35:45 <vblagoje> oh, right stupid question; he is likely to have copied some existing config from AS | |
Jul 09 17:35:49 <vblagoje> thx galderz | |
Jul 09 17:36:37 * mgencur has quit (Ping timeout: 255 seconds) | |
Jul 09 17:38:22 * bela (~bela@157-143.192-178.cust.bluewin.ch) has joined #infinispan | |
Jul 09 17:50:31 * dberindei (~dan@redhat/jboss/dberindei) has joined #infinispan | |
Jul 09 17:50:32 * ChanServ gives voice to dberindei | |
Jul 09 17:53:34 <vblagoje> galderz, the suggestion you made did not reproduce the error (everything was ok); lets wait until his comes back from vacation, provides all the details about his infinispan setup; as we test it, find the source of this issue and resolve it proeprly | |
Jul 09 17:55:22 <galderz> vblagoje, yeah, that's the sensible thing to do here | |
Jul 09 17:55:35 <galderz> vblagoje, it really should be reproducible standalone, outside AS7 | |
Jul 09 17:55:51 <vblagoje> yes, I have to see his config file | |
Jul 09 17:57:05 <vblagoje> galderz, so the cache stored value is in form of MarshalledValue in data container if storedAsBinary is true? Any other time? | |
Jul 09 17:57:28 * barreiro (~barreiro@2.82.249.97) has joined #infinispan | |
Jul 09 17:57:35 <galderz> vblagoje, yeah, MVs are used when storeAsBinary is used | |
Jul 09 17:57:40 <galderz> not any other time | |
Jul 09 17:57:51 <galderz> vblagoje, i'm pretty sure that's the key | |
Jul 09 17:57:59 * bela has quit (Quit: bela) | |
Jul 09 17:58:06 <galderz> i mean the key factor | |
Jul 09 17:59:23 <vblagoje> ok galderz |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment