Attach SOMDEV1 / DEV2 to Qumulo

[nmagee]

Reach out to RCI and discuss options around mounting specific Qumulo shares to DEV1/DEV2. This will help us consolidate pools of reference / universe data.

[nmagee]

@nsheff Alden's reply to this request was "I can do this easy". ;)

[nsheff]

@nmagee Following up on this -- is this possible yet?

I am waiting on this for #23

[nsheff]

Hi @nmagee, when you get a chance, can we follow up on this? What's the ETA? If beyond a week or so we should come up with an alternative solution I think.

[nmagee]

@nsheff -- sorry I didn't see this while I was away. Working on a proper/faster mount path, but for now this is mounted as read-only here on both nodes:

/mnt/q/shefflab

[nsheff]

I can see that; where is the corresponding mount on rivanna?
nevermind I found it.

[nsheff]

at what external URL can I see the files in that folder?

[nsheff]

I've changed LOLA around to use environment variables and point to the qumulo mount (see #50)

With this commit, I'm trying to update the docker swarm so that the mounts are now correctly pointing to the new mounts, instead of to local subfolders inside the app dir.

e936848

@vpnagraj and @nmagee am I doing this correctly?

The new config works for me locally on my computer as long as I have those docker mounts correct.

[nsheff]

How do I tell the swarm to re-create the containers with the new mount points?

[nsheff]

actually, those new mount points will only work with dev containers... so this gets tricky. can you only update dev containers?

[vpnagraj]

@nsheff i'll let @nmagee speak to the swarm config ...

but for the changes in c2ec8d1 i can confirm i was able to get things set up and running locally (with runApp() outside of docker)

[nsheff]

It seems that I somehow updated the production containers, because they're now showing the new code... I am not sure what I did.

This also means they are not working anymore.

[nsheff]

actually this has me quite confused... I did not update the master Dockerfile. I thought that was the way to update the production containers. But the production containers are currently running the code on the dev branch of LOLAweb. It does not seem like this should happen, so something went wrong either in practice, or in the way I understand things.

[vpnagraj]

weird ... i don't think this has been an issue until today ??

@nmagee is this a typo?

LOLAweb/docker/Dockerfile

Line 11 in c1fc5d8

RUN git clone -b dev https://github.com/databio/LOLAweb.git

shouldn't have the -b dev flag, right?

my guess is that we hadn't seen this (even with recent commits to dev) since the container hadn't actually been rebuilt until @nsheff did the PR and reverted it today

[vpnagraj]

ok yeah that was definitely a typo ... bb34fec fixed things so that the production container is pulling from the master branch for code (rather than dev)

[nsheff]

nice catch! 🏆 thanks -- now the production container is functioning again.

the dev container is just broken I think until @nmagee can update the swarm.

[nmagee]

Sorry @vpnagraj and @nsheff , I was away over the weekend and missed these. I'm updating the swarm today between datacenter work. I'll keep you posted.

docker-ce has been updated on the two servers, and rebooted with other regular updates.

[nsheff]

that's working now, thanks.

[vpnagraj]

this is change was included in v1.2 release