Hi, I have squid setup to log all requests for security reasons. When doing an npm install, npm sometimes gets a connection reset error and fails. In the cache log, I can see many messages like :
2022/05/12 09:20:00 kid1| helperOpenServers: Starting 1/32 'ssl_crtd' processes 2022/05/12 09:20:00 kid1| WARNING: All 32/32 ssl_crtd processes are busy. 2022/05/12 09:20:00 kid1| WARNING: 32 pending requests queued 2022/05/12 09:20:00 kid1| WARNING: Consider increasing the number of ssl_crtd processes in your config file. 2022/05/12 09:20:00 kid1| Queue overload, rejecting 2022/05/12 09:20:00 kid1| Queue overload, rejecting (with 100s of queue overloads) BUT http://www.squid-cache.org/Versions/v3/3.5/cfgman/sslcrtd_children.html says the maximum safe value is 32. Later versions don't allow values over 32. (I'm on 3.5) So, where do I go from here?? What is the "unsafe" worst case? Packet loss, high server load, squid crashing out, etc..? I did increase it to 64 and the idle to 5 so far haven't seen one of the failures from npm, but my cache log still shows : 2022/05/12 10:14:30 kid1| helperOpenServers: Starting 5/64 'ssl_crtd' processes 2022/05/12 10:14:30 kid1| Starting new ssl_crtd helpers... 2022/05/12 10:14:30 kid1| helperOpenServers: Starting 4/64 'ssl_crtd' processes 2022/05/12 10:14:44 kid1| WARNING: All 64/64 ssl_crtd processes are busy. 2022/05/12 10:14:44 kid1| WARNING: 64 pending requests queued 2022/05/12 10:14:44 kid1| WARNING: Consider increasing the number of ssl_crtd processes in your config file. 2022/05/12 10:14:44 kid1| Queue overload, rejecting 2022/05/12 10:14:44 kid1| Queue overload, rejecting Which queue do I increase to prevent the overload? Squid is configured without any auth, mostly ACL and ssl setup. Thanks _______________________________________________ squid-users mailing list squid-users@lists.squid-cache.org http://lists.squid-cache.org/listinfo/squid-users
