I noticed on Discord that my Discord Bot is not running. When I check the project on Glitch of course it starts to “install”. What does that mean? Why wasn’t it running and how would one check if it was running if checking makes it install all over again? Shouldn’t boosted mean it is running?
Another problem (probably related) but the bot code will now not install. It takes quite some time but eventually there is a log message “npm notice New major version of npm available! 7.20.6 → 8.3.0”
We don’t install npm and certainly not globally and it wouldn’t survive the next reload. So as far as I can see the project will not build and run and won’t until npm is updated.
No changes were made since the bot last built and ran by me. If this is something the version of npm is enforcing is there a command line switch to disable the update?
This isn’t good.
Small update… I let the install just proceed and it appears that it does in fact eventually build. It is now running again but that begs my first question, why wasn’t it running and how could I know it wasn’t running if checking the project makes it reinstall?
ps -A or
ps -aux or
pstree in the console should be able to tell you about whether your main node process is running or not. I feel like this issue may be due to some other project (that might be free) using up all the resources in the server hosting the container for your project.
@wh0 's project might be useful to you for looking up what projects are on the same server as yours. I’ve never seen the output of it on a boosted server so I’m curious to know whether “boosted” projects are on the same server as other boosted servers or do they share the same server as free projects.
Thanks for posting this.
What I’ve heard about Glitch is that it doesn’t projects up for long contiguous periods—we’ve seen posts saying the containers themselves only last around 12 hours. It’s something that applies to boosted and unboosted projects alike.
If any one container doesn’t stay alive long, it means Glitch needs to be able to bring up the project, sort of ‘from scratch’ in a fresh container, at any time. Again, this must be true of both boosted and unboosted projects.
If a project can’t complete its ‘install’ step, then boosting won’t make it work any better. Unfortunately. It would be worth more than $8/mo if it did though
Anyway, what about that error message? I’ve seen this “notice” that a new version of npm is out. It doesn’t mean “the old version won’t work, and we can’t continue.” I can only suggest that you investigate why it (sometimes?) fails to install or takes long (longer than it should?).
Merry Christmas and yes thanks for the console command. It does require the project to be up however and by then it is probably too late.
My research indicates that the container (server or project I don’t know at what level things are restarted) is in fact restarted multiple times a day at varying times a day. This has always happened and one would assume it happened each of the days my bot failed to start. BTW it wasn’t only my bot, what caused me to look into it was another bot that a Discord server relies on. I noticed that it wasn’t running either.
That one (seemingly) didn’t start by itself for nearly 2 weeks. I had someone view the project, it went thru the loading steps and started running. Again there is no way to see what might have stalled it via the project editor as it completed without a problem.
Would a “restart complete”, “restart fail” notification option be a good thing? It would give members some insight into when and how often a project is restarted and in 2 cases now I would have known to look into things considerably sooner.
That does sound like a good idea.
For right now, I wonder if there’s some kind of “push” based uptime monitoring, where the project periodically tells the monitor that it’s still running. It’d be kind of a hassle to need a separate service, but at least we’d know.
It sounds a little like a heartbeat but what would “running” mean in most contexts? Likely a timer pinging the monitor on a routine basis.
I considered some time ago about but didn’t post how an event system could be useful. I was thinking about a one minute warning for a system reboot for instance so steps could be taken if needed like logging or user notification. The idea that the system could simply be restarted without any notice makes it more likely a DB could be trashed, a long computation or query could be interrupted, etc. and tends to limit the apps one is likely to run on Glitch.
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.