Help! … I have deployed my app into production and the server is crashing frequently with just this in the log each time:
s2zk2 2017-11-09 00:10:40+00:00undefined
zywex 2017-11-09 00:11:36+00:00Kadira: completed instrumenting the app
zywex 2017-11-09 00:11:36+00:00Kadira: Successfully connected
So I’ve no clue what is wrong! I always deploy to a test galaxy environment first and this is working fine (as is all my local test/dev environments). I have deployed many times in the past and never had this problem. Using latest Meteor version.
Any clues as to how I proceed?
OK, I have an update on this. I have reverted to version 188.8.131.52 from 1.6 and it now works.
The only difference between my Galaxy live environment (where it was going wrong) and my Galaxy test environment (where it works) is:
a) Client connections were slightly more in live
b) Live environment is “professional” and thus has Kadira running
I don’t think it’s a). So does anyone please have any experiences with Meteor 1.6 crashing (with minimal log save for “undefined”) on a Galaxy “professional” server? Do the Meteor guys have any insight on this?
1.6 is a fairly juicy upgrade as it involves a major upgrade to NodeJs and I suspect this has something to do with it…
Um, could some-one in MDG please point me in some direction here…after all, I am paying for Galaxy and it’s not working.
I’m not looking for an answer just a suggestion as to how I can discover what the problem is. The server logs “undefined” and then crashes. So that’s not a lot to go on.
Are there other logs I can look at? Should I try switching off “professional” and see if that works? Any suggestions at all?
Have you opened a galaxy support ticket?
Thank you, I hadn’t seen you could do that!
I just noticed some similar behavior seemingly new with 1.6 that might be related. still investigating. Let me know if you get a response or find the cause!
Update. I posted a ticked 6 days ago and have since chased it up twice but I get no reply. Black hole, basically.
So unfortunately ticket posting does not work!
It’s clear that I seem to be the only one with this problem so it must be something down to my specific situation (and probably my fault) but I still have no idea how to analyse a Galaxy server crash when I only have “undefined” in the logs to go on.
Anyone have any other bright Ideas as to how I could make progress?
I think I know what this was…at least my Galaxy server is now no longer crashing.
I use many 3rd party packages (wherever possible, frankly, though I try to stick to well used ones) and I guess the more you use the more you expose yourself to incompatibilities with core upgrades. I believe my problem was due to this package:
I can’t be 100% sure and my sincere apologies to the author if this is not the case. But it seems very likely. Though it’s not really the package’s fault probably as it merely wraps this NPM package:
Trouble is I think it’s wrapping a really old version that is the likely culprit for crashing my Meteor V1.6 server. I guess using packages that merely wrap NPM packages is a bad idea now that you can use NPM proper. That said, I couldn’t get the latest NPM version to work either so I’ve ended up using another package that wraps a later version:
Which works! Though I guess I’m storing up a problem for later when some upgrade will break this one too!
Summary: Beware legacy packages and anyone out there who is using quackware:ical-generator - be warned when you upgrade to Meteor 1.6.
Yet a further update with more info…
I’m using server side routing courtesy
meteorhacks:picker. I have discovered that if you do not catch errors within the
Picker.route() function then that error is passed out to the container which duly crashes.
So now I catch all errors (and log as appropriate) and I don’t get any more server crashes.
I think it was that simple!
Can’t comment on the specific issue you had, but props for following up on your posts here with findings. Might help some others in the future. Not everyone does this. Much appreciated!