Обсуждение: buildfarm client release 4.18
I have just released buildfarm client version 4.18. In addition to some minor fixes, there are two significant changes: * The client now makes a determined effort to clean up any left over build artefacts from previous runs at the start of a run. It also tries to clean away old socket files. That means that manual fixes are much less likely to be required, if the script has crashed or the machine has crashed or restarted. * There is a new config parameter "wait_timeout", which defaults to undefined. If it is set then any run that takes longer than the specified number of seconds is aborted. That means that runs that get stuck will time out, and the whole animal won't be stuck and requiring manual intervention. My test animal has a setting of 3 * 3600, i.e. 3 hours. There is some possible danger in this, that we might miss bugs that cause us to get stuck, so at least some animals should not use this feature. The release can be downloaded from: <http://www.pgbuildfarm.org/downloads/releases/build-farm-4_18.tgz> cheers andrew
Andrew Dunstan <andrew@dunslane.net> writes: > * There is a new config parameter "wait_timeout", which defaults to > undefined. If it is set then any run that takes longer than the > specified number of seconds is aborted. That means that runs that > get stuck will time out, and the whole animal won't be stuck and > requiring manual intervention. My test animal has a setting of 3 * > 3600, i.e. 3 hours. There is some possible danger in this, that we > might miss bugs that cause us to get stuck, so at least some animals > should not use this feature. Hm, do you mean that if the timeout is exceeded, it just shuts down and doesn't report the failure? If there is a report I don't see how we'd "miss" anything. regards, tom lane
On 10/11/2016 08:51 PM, Tom Lane wrote: > Andrew Dunstan <andrew@dunslane.net> writes: >> * There is a new config parameter "wait_timeout", which defaults to >> undefined. If it is set then any run that takes longer than the >> specified number of seconds is aborted. That means that runs that >> get stuck will time out, and the whole animal won't be stuck and >> requiring manual intervention. My test animal has a setting of 3 * >> 3600, i.e. 3 hours. There is some possible danger in this, that we >> might miss bugs that cause us to get stuck, so at least some animals >> should not use this feature. > Hm, do you mean that if the timeout is exceeded, it just shuts down and > doesn't report the failure? If there is a report I don't see how we'd > "miss" anything. > > Yeah, unfortunately. I can look at making it report something, not sure exactly how much will be possible. cheers andrew
Andrew Dunstan <andrew@dunslane.net> writes: > * There is a new config parameter "wait_timeout", which defaults to > undefined. If it is set then any run that takes longer than the > specified number of seconds is aborted. That means that runs that > get stuck will time out, and the whole animal won't be stuck and > requiring manual intervention. My test animal has a setting of 3 * > 3600, i.e. 3 hours. There is some possible danger in this, that we > might miss bugs that cause us to get stuck, so at least some animals > should not use this feature. Hm, do you mean that if the timeout is exceeded, it just shuts down and doesn't report the failure? If there is a report I don't see how we'd "miss" anything. regards, tom lane
On 10/11/2016 08:51 PM, Tom Lane wrote: > Andrew Dunstan <andrew@dunslane.net> writes: >> * There is a new config parameter "wait_timeout", which defaults to >> undefined. If it is set then any run that takes longer than the >> specified number of seconds is aborted. That means that runs that >> get stuck will time out, and the whole animal won't be stuck and >> requiring manual intervention. My test animal has a setting of 3 * >> 3600, i.e. 3 hours. There is some possible danger in this, that we >> might miss bugs that cause us to get stuck, so at least some animals >> should not use this feature. > Hm, do you mean that if the timeout is exceeded, it just shuts down and > doesn't report the failure? If there is a report I don't see how we'd > "miss" anything. > > Yeah, unfortunately. I can look at making it report something, not sure exactly how much will be possible. cheers andrew