Omnimaga
Omnimaga => Discontinued => Our Projects => OmnomIRC Development => Topic started by: DJ Omnimaga on November 21, 2012, 03:24:55 pm
-
I started haivng some issues with Omnom lately:
-When I try to send text, nothing seems to happen. The text remains there and I can't type anything anymore. However, if I refresh the page, the text actually sent. I waited minutes sometimes, to see if it was just lag, but nothing happened. It happens every single text sending attempt.
-Because of this, I am forced to refresh the page everytime I sent a line of text, which can be annoying during long OmnomIRC convos.
-When you switch to a different OmnomIRC channel tab, it remains blank forever, requiring a refresh.
-
What browser?
Have you tested in other browsers?
Can you see any errors on the console for the browser which is having this issue?
Have you tried clearing your cache/hard refreshing the page?
-
I was having problems too except it just lags when I click enter. I have to wait until it sends it or at least shows that it was sent.
I'm using Firefox 16.0.2.
-
That happens in all browsers Nicco. At least it I've never seen anyone not have lag
-
What browser?
Have you tested in other browsers?
Can you see any errors on the console for the browser which is having this issue?
Have you tried clearing your cache/hard refreshing the page?
Opera, but I haven't checked any console (Idk how to do that in Opera lol) and I didn,t have time to check other browsers.
It started happening a few days ago, but it just got worse and worse to the point where even after 1 minute it wouldn't have sent. I tried hard-refreshing the page, to no avail.
-
It also happens in chrome but it 'only' freezes my page for ~10 seconds. I have the impression that it happens only when there's a lot of traffic.
-
http://help.opera.com/Linux/9.52/en/devtools.html <- How to open the console in Opera.
https://developers.google.com/chrome-developer-tools/docs/overview#access <- Open console in Chrome.
Ctrl-shift-k <- open console in firefox for windows/linux Mac is (apple key)+shift_+k. You can also access it under tools->Web Developer->Web Console or Firefox Menu->Web Developer-> Web Console.
Everybody who reported this should check their console and see if there is anything abnormal going on. Post a log of all the errors/warnings they see. I'm wondering if it might have to do with omnomIRC not being hosted on omnimaga anymore.
-
I tried that and all I get is
Failed to load resource http://www.omnimaga.org/mobiquo/tapatalkdetect.js
2Blur Omnom_Parser.js:858
2Focus Omnom_Parser.js:859
2Blur Omnom_Parser.js:858
-
Looks like there is an issue with tapatalk on non-mobile browsers. That is unrelated though. The blur/focus things are just it logging when you don't focus and then focus again on the window. No errors with Omnom at all. It must be network lag. Has anybody watched the network panel when they send a message through Omnom and seen if anything strange happens?
-
Eeems Alberthrocks tends to use oversellers as VPS, as in the ridiculously cheap ones that will give several GB of space and bandwidth for like $5 instead of $35, but impose invisible limits on customers, along with having lower specs. Omnom is very resource intensive, so it might just be using his server resources at the limit or maybe the VPS internet connection is just crappy. I didn't have time to check the console, though, so I'm gonna check that later.
Btw Eeems your keyboard shortcut actually asks me if I want to create an Opera e-mail account O.O
-
Eeems Alberthrocks tends to use oversellers as VPS, as in the ridiculously cheap ones that will give several GB of space and bandwidth for like $5 instead of $35, but impose invisible limits on customers, along with having lower specs. Omnom is very resource intensive, so it might just be using his server resources at the limit or maybe the VPS internet connection is just crappy. I didn't have time to check the console, though, so I'm gonna check that later.
Btw Eeems your keyboard shortcut actually asks me if I want to create an Opera e-mail account O.O
S3rv3r 1 (main s3rv3r):
K|V|M V1rtual1zat1on (no overselling!)
Intel(R) Core(TM) i7-2600 CPU @ 3.4 GHz, 2 cores
1 GB RAM
40 GB HDD
1OO MB1t/s n3tw0rk
Debian 6 32-bit (when Debian 7 is released, will be upgraded to Debian 7 64-bit)
Nope. Not all of them are oversellers ;)
I'm very aware of how all of his servers work, since I am an admin for them and regularly work on them. Plus, Omnom is nowhere as resource intensive as it use to be, plus we have things on our server that help keep things from using up all the cpu.
What keyboard shortcut?
-
Ehh.... actually, the server running this is server #2, aka:
S3rv3r 2 (secondary s3rv3r):
0penVZ V1rtual1zat1on (overselling possible)
Not as reliable! (But very well spec'ed!)
Intel(R) Xeon(R) CPU L5410 @ 2.33GHz, 4 cores
2 GB RAM
50 GB HDD
1OO MB1t/s n3tw0rk
Debian 6 32-bit (when Debian 7 is released, will be upgraded to Debian 7 64-bit)
(This is backup2.withg.org)
I'm not 100% sure if the owner is actually overselling or not, but OpenVZ is a software known for being able to have tons of VMs running on a server that can barely handle them. The price is also rather low when I made my purchase - $7/month for life. The VPS provider doesn't exactly have the greatest reputation, but...**
As for CPU usage, CPUHero is not running on backup2 since OmnomIRC is running, and murder might occur if CPUHero runs :P I've since implemented an exclusion function, but I haven't installed CPUHero on the server OmnomIRC's on yet.
The only thing that could cause high CPU usage is the Minecraft server (now decommissioned since no one uses it), and the backup daemon (which will be turned off very soon). However, there are 3 cores left to use (since the backup daemon isn't multi-core), so speed shouldn't be a problem.
The only thing that I might do (if necessary) is implement some sort of restarting service (like the Newy one), where you can click Restart to restart OmnomIRC. I would probably require 5 or more people, though, considering that it's a major service restart. (I would be restarting NGINX, PHP5-FPM, MySQL, and OmnomIRC itself.)
Again, backups on that server will be disabled - they only existed to grab a few things off the server before it died. Now that I've grabbed those things, backups will be disabled - they cost money! ($5/month + storage costs)
** Their server got hacked, and their VMs were being deleted pretty rapidly, hence why my interest in backing up spiked.
EDIT: Heh... I do have something running, but it's CPULimit, and it's set to limit OmnomIRC :P CPUHero would do the opposite - limit other CPU hungry processes in favor of OmnomIRC.
Also, the only error I've seen seems to have happened around the time this topic was made:
2012/11/21 20:19:34 [error] 14004#0: *3584158 upstream prematurely closed connection while reading response header from upstream, client: IP_REMOVED, server: omnomirc.withg.org, request: "GET /Update.php?lineNum=REMOVED&channel=REMOVED, HTTP/1.1", upstream: "fastcgi://127.0.0.1:9000", host: "omnomirc.www.omnimaga.org", referrer: "http://omnomirc.www.omnimaga.org/"...which meant PHP5 died during data send, or something wonky occurred with NGINX<->PHP5-FPM communication. But since then, no errors.
-
Eeems Alberthrocks tends to use oversellers as VPS, as in the ridiculously cheap ones that will give several GB of space and bandwidth for like $5 instead of $35, but impose invisible limits on customers, along with having lower specs. Omnom is very resource intensive, so it might just be using his server resources at the limit or maybe the VPS internet connection is just crappy. I didn't have time to check the console, though, so I'm gonna check that later.
Btw Eeems your keyboard shortcut actually asks me if I want to create an Opera e-mail account O.O
S3rv3r 1 (main s3rv3r):
K|V|M V1rtual1zat1on (no overselling!)
Intel(R) Core(TM) i7-2600 CPU @ 3.4 GHz, 2 cores
1 GB RAM
40 GB HDD
1OO MB1t/s n3tw0rk
Debian 6 32-bit (when Debian 7 is released, will be upgraded to Debian 7 64-bit)
Nope. Not all of them are oversellers ;)
I'm very aware of how all of his servers work, since I am an admin for them and regularly work on them. Plus, Omnom is nowhere as resource intensive as it use to be, plus we have things on our server that help keep things from using up all the cpu.
What keyboard shortcut?
The keyboard shortcut in your previous post (CTRL+SHIT+K). And I see.
EDIT: Wow the CODE tags are so messed up in tapatalk! o.o
-
The keyboard shortcut in your previous post (CTRL+SHIT+K). And I see.
That keyboard shortcut is for firefox ;)
-
Oh ok lol, I thought it was for Opera since you know I use Opera ;D
-
Ctrl-shift-k <- open console in firefox for windows/linux Mac is (apple key)+shift_+k. You can also access it under tools->Web Developer->Web Console or Firefox Menu->Web Developer-> Web Console.
I mentioned it was for firefox in the post ;P
-
Oh right I didn't see it, because you didn't put a link first like the 2 first lines, and I thought it was continuing on the other lines. :P
Anyway cookie/cache clearing + hard refresh + restart didn't solve the problem, but for some reasons it doesn't seem as bad since yesterday.
-
Ah lol, well I wasn't.
Did you chck the console for network traffic to see if there was anything weird?
-
I didn't since someone posted error logs the other day but I might later if I notice the lag starts again. (Seems to be mainly during evening or night now, while Friday it was 24/7)
-
They weren't actually error logs. It was just normal operation. I'm looking for the networking tab, to see the stats on the transfers etc.
-
The extreme lag has gone from worse to something I cannot describe now. It stops responding to any text sending requests for about 30 seconds, then suddenly processes all of them for 1-2 seconds before stopping responding again for 30 secs, and so on.
Also, since the past 3 days, when clicking the back button, OmnomIRC stops working until next refresh.
-
Earlier I was using Chrome and it would freeze the entire page attempting to send a message.
By the looks of it, it's doing it again now as well. It was fixed the first time when Omnom was apparently disconnected and was reconnected.
-
I just checked at work. Same thing happens to me. I think it may have to do with network lag. The strange thing is there is hardly any lag between the servers right now, so no idea why it's acting up.
-
Some parts of Firefox fail to respond completely, like pressing New Tab or closing the Find bar. That... shouldn't be happening.
-
Earlier I was using Chrome and it would freeze the entire page attempting to send a message.
Same here :o
-
That's actually something that happens with xhr requests taking too long. I noticed that as well. There are ways to get around it though. The fact that there is this much lag is confusing me though. The ping between the servers is ~50ms
-
The weird thing is how it's intermittent as well. (eg how it suddenly starts responding to requests after 28 seconds or so and 2-5 seconds later it stops again)
Is there anything else than OmnomIRC running on the server like Minecraft or random people scripts that might not necessarily be coded properly (I remember alberthrocks offering hosting, right?)
-
Only me an albert use this specific server at the moment. Minecraft was shut down a while ago, but while they both were running there were no issues. I took a look at the server, and all 4 cores are operating under 10% and RAm is well within acceptable bounds. The issue is not with the server. It may be with cloudflare. I'd have to go home and do some more digging. That wont happen tonight though.
-
Do you mean CloudFlare could be trying to block OmnomIRC connections? I know it tends to block lots of unnecessary traffic from bots and crawlers, and OmnomIRC happens to not be a human person.
-
No, I'm not saying that. I'm saying cloudflare might be causeing some strange routing that makes the data take forever to get around. Cloudflare wouldn't try to block traffic from leaving the server, only traffic coming in that we don't allow.
-
What I noticed when this happens (a bit less frequently lately) is that if I refresh the page, my text is already displaying in the channel.
Another thing is how when this happens is that the OmnomIRC scrolling arrows no longer works until the text has sent or until I refresh the page.
-
It also happend more often recently that omnomirc lost the connection
-
Do you mean the same thing as my problem or an actual error message? I think I got one once, but it was long ago.
-
i mean the actual bot connecting efnet/omninet/omnomirc
-
Oh ok. Often that is EFNet or Omninet fault, though.
-
The scrolling issue would be the javascript connection to the bot blocking all other javascript until it is done working. There might be a way to code around that. I'd have to look at how it's coded.
The bot losing connection might have to do with the server it's on. I'd have to look into it, or get Albert to look into it.