-
Notifications
You must be signed in to change notification settings - Fork 4.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Kibana docker hangs our test on travis and jenkins #5953
Comments
@ruflin could you take a look? |
@ph I think this one is out of our control. It's potentially only indirectly related to #5928 The reasons this times out (especially on travis) is because the optimize process takes longer then 10 minutes on a machine with not too much processing power. I also see this on my local laptop from time to time. The problem could be circumvented if there would be an option that during the optimize some logs are outputed but there is no option for that in Kibana. I would expect it to work on Jenkins except if Kibana snapshot is broken (which can happen from time to time) but is a good way to figure out issues early. I tried the most recent version locally and it seemed to build Kibana as expected (but it takes a long time). I will flush all images again and retry. The way I'm hoping to solve this problem is that we will have the docker images for snapshots also available somewhere in a registry, meaning we don't have to build it ourself each time. This is potentially going to happen soon ;-) |
On one of the most recent jenkins builds I got an output I haven't seen so far:
@elastic/kibana-operations Did anything recently change in Kibana which now requieres quite a bit more memory for the optimize step? The above is based on master / snapshot. |
@elasticdog I'm kind of surprised to see the above as the jenkins slaves have more the 1.5GB of memory. Is there a limitation on how much memory each container can consume on our Jenkins slaves? |
I'm wondering if it could be solved by using one of the flags here on the Kibana side: facebook/react-native#12686 But that does not explain what it builds on my machine but not (anymore) on the Jenkins worker. |
@ruflin I believe someone pinged you on slack, but for posterity Kibana's right at the file limit for production bundling and we're starting to see it fail. We're working on it now and tracking at elastic/kibana#15683. You're spot on with the flags, depending on how Kibana's being ran you'll want to either set the NODE_OPTIONS=--max_old_space_size=4096 or pass it in as an arugment to node, |
* Add `NODE_OPTIONS="--max-old-space-size=4096"` to circumvent elastic/kibana#15683 * Make it more obvious in the Docker file which parts are copied over and which are from the official repo * Remove xpack.monitoring config option as this seems to cause a rebuild. * Increase LS timeout for health check * Remove duplicated kafka health check * Increase health check timeouts to 10min * Set defaults for `ARGS` in kibana container. This makes it easy to also build the container directly. * Adjusts docker image paths to all use `ES_BEATS` variable. * Remove `detector_rules` for ML as not supported / required anymore in master. * Remove unused `SHIELD=false` variable in compose file * Fix geoip looksup that changed and improve error logging It would be great if we could disable most of the Kibana x-pack plugins as they are not needed for our testing. But the problem is 1) we cannot install only 1 plugin AFAIK 2) disabling one of the plugins causes a rebuild which makes the build even longer. Closes elastic#5953
* Add `NODE_OPTIONS="--max-old-space-size=4096"` to circumvent elastic/kibana#15683 * Make it more obvious in the Docker file which parts are copied over and which are from the official repo * Remove xpack.monitoring config option as this seems to cause a rebuild. * Increase LS timeout for health check * Remove duplicated kafka health check * Increase health check timeouts to 10min * Set defaults for `ARGS` in kibana container. This makes it easy to also build the container directly. * Adjusts docker image paths to all use `ES_BEATS` variable. * Remove `detector_rules` for ML as not supported / required anymore in master. * Remove unused `SHIELD=false` variable in compose file * Fix geoip looksup that changed and improve error logging It would be great if we could disable most of the Kibana x-pack plugins as they are not needed for our testing. But the problem is 1) we cannot install only 1 plugin AFAIK 2) disabling one of the plugins causes a rebuild which makes the build even longer. Closes #5953
We have been noticing a few hangs in our test suite which seems to come from the merge of the Kibana docker changes in #5928
The text was updated successfully, but these errors were encountered: