GSA+CS crashing servers?
We recently have some weird issues with either GSA or Captcha Sniper (or both) on our private servers.
We found that our VPS and our Dedi Servers crash over night. They seem to be running, but we can't connect via RDP. After talking to PowerUp (one of our providers and well recommended), he gave us some insight and the issue may have to do with the use of unused RAM resources, subsequently resulting in a system crash - but we are not 100% sure.
We had this issue over more than 5 servers for the last 4-5 days. We are sure we are not the only ones with that problem, but it's rather frustrating to restart servers all the time.
Would be great if any of you could confirm our problem.
Setup:
330 Proxies per server
~200-300 threads per server
CSX
No scraping, just using our own link database.
Cheers,
mmtj
Comments
Unable to access the server via remote access in the AM.
GSA is crashing my server for the past 2 nights.
It has never done this before. 2 days ago i upgraded from 4.58 -> 4.61. By this morning i could not access the remote server which meant it was hung up or crashed. I had to do a reset for the 2nd time.
Finally when i logged in GSA stated it was back at v4.58? How is that possible? I did an upgrade 2 days ago. Then i ran though the upgrade again.
Whats going on with this?
Also, you state this in the changelog; 4.60 - new: added a basic default badword filter list that gets added to newly created projects. You still need to enable that option though.
But where is this option? When i click NEW i dont see this option anywhere.
Thinking about it, i should not have upgraded to 4.61 again this morning. I should have left it alone and continued to run 4.58 until such time that this issue is fixed. I have no idea how it reverted back after this crash anyway.
For testing purpose, I setup a brand new server, I decided to run only 1 Project @300 threads on the dedicated server. Here are the initial stats:
CPU : 13%
RAM: 23%
I left them running for 24 hours and then the consumption of the resources increased after 24 hours. Here are the stats:
CPU : 15%
RAM: 50%
I decided to leave them running for another 24 hours and the RAM consumption further increased. Here are the stats:
CPU: 15%
RAM: 80%
After this, I decided to close both the programs one by one. Here are the stats after closing GSA.
CPU: 15%
RAM: 80%
So with only CS running (it wasn't even breaking any captchas), above were the stats. So based on these tests, I can make an assumption that GSA is leaving a bad port open for CS which is resulting into a memory leak issue and causing servers to crash.
I have PMed Sven about this issue and I can setup some more test servers for him to ensure that issue gets fixed very soon.
Either way, we gotta find out the root cause and get it fixed
I'm thinking maybe a batch file to kill and restart CS, or something like that.
I had GSA running the whole night together with a "captcha-solving-app" (other than CS).
No memory leaks on both sides. All running stable without any issues. So as CS was not involved here I can only assume that CS is the problem. Just a guess though.
I can't run CS on my system as it keeps crashing almost immediately (tessact error). Maybe I should give it a try with that popup closing app that Ozz mentioned once.
http://www.microsoft.com/resources/documentation/windows/xp/all/proddocs/en-us/agent_add_task.mspx?mfr=true
You can Schedule a restart task for your VPS/Server inside Windows every 6-12 hours. This is a nice lil decent solution until the everything gets sorted out.
Well I got this tesseract pop-up almost on every captcha that is analysed by CS. And that was from the start. Only solution for me was the exchange of the exe with a newer version. But even that was crashing sometimes. But you mentioned once that this is not a good idea as you tested CS only with that older one and so I didn't run CS anymore.
A test with just exchanging CS with another captcha-breaking-executable was not showing any errors either crash or memory leaks as Ozz pointed out as well. And the way GSA-SER is sending the captchas is the same here.
@CaptchaSniper naw don't get me wrong, I only tried to see what the problem is. As I can't run CS for some odd reason, I was using another one and as this one communicate in a same way with SER than CS and everything was OK, I thought I would through it in here to get this sorted out.
Anyway good you got it fixed on your end
@mozzer, thats right. The program will wait till it gets an answer from CS before sending a new captcha to analyse. Thats done to not cause any conflicts or hammering CS.
This however is not slowing things down in the end as CS can only work on one captcha at a time and would queue things anyway (Im not sure if it does though). You might see threads hanging but thats due to the wait of CS to answer the captchas.
But yesterday i was thinking about something curious, i don't know this for a fact, but i think CS is older than GSA SER, and is used for a lot more software than just SER, do other programs also have these compatibility issues using CS ?
seems the problem is from CS because when i closed CS the memory usage dropped to 45% , when it was around 91%
i think the CS is causing the server crashes , and hopefully this gets fixed soon !
Do you send eMail to your user base when an update of CS is available ?
Because I don't see an option on CS for an update check.
regards...
backup the CS program folder just in case.
extract the zip file
copy over and replace the exe file in the CS program folder.
that's it.
Thanks for the update. Lets see how this goes.
Now, every time i log in to my server i see the Tesseract box open. This used to happen sometimes and not always on Win2008 R2. Its always happening now. And it randomly appears daily, more than 3 times (Arrrggggggggggg) a day.
So this annoying thing is still around. Please dont ask me to run another program to counter act it. I ll simply switch back to the old version regardless of the memory leak. My updated consisted of a rename so i can still do that without a problem.
+1 here for annoying Tesseract.exe error (Windows 7 server), problem is that you can't rely on software to run overnight, during the day I do check it but when I leave it to work unmonitored it pops up that error and after a while whole thing crashes...
Anyone face the same problem?
@mikie46 I agree. I now have 5X the number of Tesseract errors and things slowed down, so I went back and replaced the old .exe file back in the CS folder. I haven't been experiencing the leakage issue like some of the others, so for me I will wait until the fix gets a little better tuned up by CS.
@darman82 I saw that I was getting a bunch of Terreract errors and things weren't running as smoothly, so I shut everything down and replaced the old .exe file. I sensed I was heading in the direction you described, and decided to roll back.
but with how much it is freezing lately is making it very hard for me to run, I have 7 servers so I have to check each one like 3-4 times a day. Does anyone have a link to the old .exe i'd like to roll back as well. I'd rather put up with the tesseract errors than it freezing up all the time.
After running Fixed patch, it crashes a lot no matter how many threads I put in, I tried with like 17 threads and so... It's all the same... Now after testing:
I do have to recommend something to all users, I think there is absolutely no point of setting CS to do retries if Captcha is wrong, as I think that effects stability of the software and I'm not even sure if GSA knows that captcha that CS sent is wrong or right...
Hm, does this even makes sense? As I've read somewhere that CS has no idea if captcha is right or wrong... So how does GSA knows?
In that case does GSA actually keeps sending SAME captcha to CS to read it again for 3 times (or how many retries we set)? Does this makes sense?
Just my observation on the issue...