can't access web interface after deploying several snapins

[fogusnew]

Dear fog community,
until now everything worked really well for several months.
I’m running fog server 1.5.4 on a simple debian 9 32bit box (via .sh script install). Imaging Windows and also sending snapins worked most of the times.

Today I started the deployment of an office 2016 snapin to ~25 pcs (win10). Since this the web interface of the fog server went down and I can’t access it anymore (error 503). As far as I can see apache, php and mariadb is running.

The deployment did not finish yet (I guess), but - obviously - I can’t see the status… I just can’t access the interface.

Can anybody point me in the right direction to solve this issue?

Thanks so far.
Chris

[Sebastian Roth]

@fogusnew Is the whole MS Office 2016 installer within the snapin or is it just a script to install it from a UNC path? If you have the whole Office installer uploaded as snapin I can imagine this causing the web UI to fail as the clients pull the whole binary from the server. In case the download fails (for whatever reason) the clients will try over and over again.

shell> mysql -u root -p
Password:
...
mysql> use fog;
...
mysql> UPDATE snapinTasks SET stState=5;
...
mysql> UPDATE snapinJobs SET sjStateID=5;
...
mysql> UPDATE tasks SET taskStateID=5 WHERE taskTypeID=13 OR taskTypeID=12;
...

This should mark all snapins as finished and a couple minutes later your web UI should be usable again.

[fogusnew]

what I found - and maybe it is helpful to know - is that apache runs on about 30 instances. systemctl status apache2 gives me two pages of pids. I restarted the server a couple of times - same picture. Is this “normal”?

[Sebastian Roth]

@fogusnew Is the whole MS Office 2016 installer within the snapin or is it just a script to install it from a UNC path? If you have the whole Office installer uploaded as snapin I can imagine this causing the web UI to fail as the clients pull the whole binary from the server. In case the download fails (for whatever reason) the clients will try over and over again.

shell> mysql -u root -p
Password:
...
mysql> use fog;
...
mysql> UPDATE snapinTasks SET stState=5;
...
mysql> UPDATE snapinJobs SET sjStateID=5;
...
mysql> UPDATE tasks SET taskStateID=5 WHERE taskTypeID=13 OR taskTypeID=12;
...

This should mark all snapins as finished and a couple minutes later your web UI should be usable again.

[fogusnew]

@Sebastian-Roth
Well, as you mentioned, it’s the whole installer. I deployed it successfully to ~5 PCs several weeks ago, so I did not think of the consequences when starting a deployment to 27 boxes.
I’m off site now but I will try to apply the suggested solution as soon as I’m at the place again.
I’ll report back then.
Thanks a lot so far!

Chris

[fogusnew]

@Sebastian-Roth Thanks for your help! I visited the site today only to discover that the webinterface was perfectly accessible. I’ve done nothing.
So that lets me think this was / is a performance issue. The machine the fog server is running on is not a big one, so it’s very likely that this computer performs bad.
However. I also discovered that the snapin was installed on a great number of machines, but unfortunately not on every box. And the GUI tells me that many snapin tasks are still “queued”, even though the software already works on some of the associated clients.

Might this all be related to this weak server machine? I would like to go into details for I really like the fog system and plan to implement it on at least two more sites I work at. Obviously I’d like it to be reliable.
I guess I should open another issue?

[Sebastian Roth]

@fogusnew I am sorry to say this but FOG is way more reliable in terms of imaging than it is when it comes to deploying snapins. I think we’d need to rework that part of the code to make it more robust. But I don’t see this happening any time soon as development power is scarce at the moment.

Not sure what to advise… Help us working on FOG and we might get the snapin part better.

[fogusnew]

@Sebastian-Roth
Ok, I see. Time is limited! I totally understand this. Well, I’d like to help but I’m have no developing skills at all. I’l try to further investigate the problem (saw some errors in the fog.log)
For the moment this issue can be closed.