I was checking out some of the recent failures and found that the server wasn't starting by the timeout. DBT2 shouldn't have been failing after the secure-file-priv fix, which helped me troubleshoot.
The code in server/drizzled.py's start() method includes a timeout, but it wasn't making sure that the server was actually running by the end of that time. I have added code to the start method that will print an error message and return an error code to the test. I've also increased the timeout values by 10s. It would be interesting to know if the server itself has slowed for startup or if the machines are slow, but that is the root cause of all of the latest batch of drizzleslap/dbt2 failures. On Wed, Jun 23, 2010 at 7:45 PM, Lee Bieber <[email protected]> wrote: > I would be interested to know what changed that would have caused this new > behavior? Everything was working fine until recently so it would be good to > know. > > -Lee > > On Tue, Jun 22, 2010 at 3:55 PM, Patrick Crews <[email protected]> wrote: > >> Hi everyone, >> >> Just wanted to provide an update on DBT2 and Drizzleslap. >> >> We had been seeing a lot of failures for these tests and I now believe >> they have been sorted out. Both of these tests were failing early on >> (initial server startup / pre-test-data loading) >> >> For DBT2, it was that we weren't specifying a secure-file-priv value so we >> could load the DBT2 data (Thanks to Lee for helping to point me in the right >> direction here). It has been tested and works fine now. The config file >> now includes the secure-file-priv option and this has been pushed to trunk. >> >> For Drizzleslap, I will admit that I am not 100% certain what fixed this. >> As randgen used a similar process for starting / setting up the servers, I >> included some calls that randgen/run.py used that weren't in >> drizzleslap/run.py (mostly with setting up startup options for the server). >> I don't believe this fixed things, but it could be possible that I tested >> too quickly. The last thing I did was to add a call to server.ping() after >> the initial server.start(). >> >> From looking at the code in server/drizzled.py, it seems that we start the >> server and wait a given amount of time, but we don't really fail the test if >> the server hasn't started by the end of the timeout. I added the call to >> ping() to make sure the server is really running before proceeding. >> >> It could be possible that the server was a bit slow to start up or my >> addition of the code from randgen fixed things. I'm not sure. >> >> In any case, drizzleslap is currently running, but we'll need to keep an >> eye on things for a bit to make sure that it works consistently. >> >> Thanks, >> Patrick >> >> _______________________________________________ >> Mailing list: >> https://launchpad.net/~drizzle-builds<https://launchpad.net/%7Edrizzle-builds> >> Post to : [email protected] >> Unsubscribe : >> https://launchpad.net/~drizzle-builds<https://launchpad.net/%7Edrizzle-builds> >> More help : https://help.launchpad.net/ListHelp >> >> > > _______________________________________________ > Mailing list: https://launchpad.net/~drizzle-builds > Post to : [email protected] > Unsubscribe : https://launchpad.net/~drizzle-builds > More help : https://help.launchpad.net/ListHelp > >
_______________________________________________ Mailing list: https://launchpad.net/~drizzle-builds Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-builds More help : https://help.launchpad.net/ListHelp
