Bug 987750
| Summary: | Exceeded timeout installing CouchDB | ||
|---|---|---|---|
| Product: | OpenShift Online | Reporter: | David Perrales <perales> |
| Component: | Containers | Assignee: | Hiro Asari <hasari> |
| Status: | CLOSED UPSTREAM | QA Contact: | libra bugs <libra-bugs> |
| Severity: | high | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 1.x | CC: | hasari, jkeck, matzew, perales |
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2013-07-25 17:54:36 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
|
Description
David Perrales
2013-07-24 05:36:42 UTC
Do you know if this quickstart ever worked without timing out? We now have a timeout of 3600 seconds on shell command execution. Increasing the timeout may solve the problem, but doing so could affect other gears sharing the node. (I do not believe that the timeout can be altered by the quickstart.) This quickstart downloads (in pre-build) and compiles (in build) from source many software packages, which is the problem. The pre-build hook downloads the archives on EVERY code push. Downloading time eats into the time allocated for deployment. You *could* add the archive to the git repo, but that would add bloat to the repo itself. You *may* be able to break the process into pieces; for example, on the first 'git push', download and build ICU. On the next push, download and build spidermonkey, and so on. This may work, if no single step exceeds the timeout, but there is no guarantee. (I am particularly concerned with downloading and compiling Erlang.) Ideally, Erlang and CouchDB should be supported cartridges, but I do not think it is on our radar. I advise you to propose it at https://www.openshift.com/ideas. As a benchmark, I tried using this quickstart (without any alteration) on a development image with nothing else on it. It was able to complete deployment (but barely-—maybe in 58 minutes). You are right. There are a lot steps and is too long process. I will try to resolve this problem finding an alternative solution. We finished our private openshift installation. I will change timeout value in a node and try to install this quickstart. After that is simple to upload the result to another repo or use gear snapshot capability. Anyway thanks for your time. David, Just a thought... If you have control over the nodes that your gears run on, you could bypass a lot of the hassle by installing (manually or otherwise) couchdb package (which will install Erlang as a dependency). That is a good idea but the final objective is deploy this gear into the public paas. I need to resolve this issue in a different way. Getting the remote: Error message: Shell command '/var/lib/openshift/526975de50044650db0000e8/app-root/runtime/repo/.openshift/action_hooks/build' exceeded timeout of 3419 Error as well :-( The current process to make CouchDB (via an action hook) is too time-consuming. We will need a real cartridge that installs the 'couchdb' RPM package. Looks like we have version 1.0.4 available in the repository. I created https://trello.com/c/Iid6N299/322-create-a-couchdb-cartridge which will be prioritized and triaged in due time. |