0

We have a cloud server. We have load balancer of 3 server and one public IP. We have master slave for DB Connection.

The application serves some data to users, the users post some data and some files to server. Currently the scenario is when the number of users uploading files increases the whole system slows down and many users are not able to perform the basic operation in the web app. the CPU usage is below 25%.No error logs on the application end. The resource are free. The only thing we got is bandwidth getting choked when there are many users uploading files. Is there any solution where we can exactly detect error logs or why it is slow?

Akki
  • 3
  • 2

1 Answers1

0

It is most likely that the cloud provider throttles your network I/O. You should check the monitoring tools of said provider and create an alert when you reach certain limits. Or maybe there was already an alert, but no one checked it (yet)?

Erik Norman
  • 86
  • 3
  • We recently monitored using the iptraf tool where we found that the throughput is going beyond 150mbps but the provider said that we are being assigned 100 mbps is this the issue of server slowdown and application lagging – Akki Oct 20 '21 at 12:49
  • I guess so. If the provider says you are hitting > 150 mps but your resource only has allocated 100 mps, then that should result in throttling. – Erik Norman Oct 21 '21 at 09:15
  • Hi Erik we updated the Server throughput to 1gb but still faced the same issue of server working slow ... One more thing is when the users increases, the website opens quickly but when the user clicks the login page and a http post request is sent to server with a photo it takes around 10 seconds to reach the nginx server from client – Akki Oct 21 '21 at 09:35
  • Phew. Difficult without knowing your exact setup. Do you have a test instance where you could do some load-testing and extensive logging? The logging - with timestamp - would help checking where the bottleneck is... – Erik Norman Oct 21 '21 at 10:11
  • 1
    Hi Erik, I think i found the glitch , we were accessing some of the images using scandir php function from around thousands of folders which was working too slow and was choking the parallel file process!! Thanks for offering the help!! – Akki Oct 25 '21 at 11:41
  • Glad you fixed the problem, and curious how you detected it... Did you add some logging? – Erik Norman Oct 25 '21 at 12:06
  • Our portal have two client end one is for users and one for the admins ,the admins used to see the images of the users from the admin portal . We thought that the error was at the user end as the images were getting uploaded from that end ,we almost went through each line but got nothing ,then we checked the whole system from admin to users all the places where files were involved in the scenarios whether read or update, then we found that while the admin is fetching the images from the users thats the only scenario where the scan dir is used and that chocked the whole queue... – Akki Oct 27 '21 at 09:16
  • Nice, so we can mark this issue as solved? :) – Erik Norman Oct 28 '21 at 05:56