Should possibly create the tmpdir by means other than creating it in source about sls-packaging HOT 8 CLOSED

Yeah, we can do something different here.

from sls-packaging.

For Cassandra and ES you also need to set jna.tmpdir (similar to java.io.tmpdir), but might want them to be different, in which case you'd need to also create that a directory that is readable, writable, and executable (and isn't mounted with noexec as some /tmp/ or /var/tmp/ may be).

@tpetracca for SA

https://docs.datastax.com/en/landing_page/doc/landing_page/troubleshooting/DSE/noWriteTmpDir.html and elastic/elasticsearch#18272 for background

from sls-packaging.

It's not clear to me that gjd should create var/data and var/data/tmp at all?
If we do create them, then do we agree that they should always be empty?

(this was uschi2000/robert with a different account)

from sls-packaging.

Not sure what the state of SLS spec is, but I'd propose that GJD should create the minimum expected directory structure with appropriate permissioning, to include var/data/ and var/tmp/ (to be consistent with Linux filesystem hierarchy standards http://www.tldp.org/LDP/Linux-Filesystem-Hierarchy/html/var.html )

from sls-packaging.

Hmm. I think it's generally safer to be explicit in your application: if you're expecting to write to var/data then you should make sure the directory exist and optionally create it if not. var is guaranteed to exist as per SLS.

from sls-packaging.

@uschi2000 you have to choose between either not setting java.io.tmpdir or making sure that the directory you set it to exists. Otherwise when any part of java tries to create a temporary file, it'll crash.

from sls-packaging.

Ah, yes I forgot about the tmpdir setting. Cool you're right.

On Sun, Sep 11, 2016, 06:13 James Baker [email protected] wrote:

@uschi2000 https://github.com/uschi2000 you have to choose between
either not setting java.io.tmpdir or making sure that the directory you set
it to exists. Otherwise when any part of java tries to create a temporary
file, it'll crash.

—
You are receiving this because you commented.

Reply to this email directly, view it on GitHub
#114 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AS4-yp5Cmxm7eYdxzUka2fuFcmYGtAkMks5qo_6TgaJpZM4J2r_I
.

from sls-packaging.

fixed by #119

from sls-packaging.