I am troubleshooting intermittent latency on a server,  several processes will report slow action at the same time every now and then,

something odd it that some of the actions are most likely not using any mutexes, I can not imagine how they can end up blocking unless the process is actually suspended by the OS, however the CPU usage and load average is quite low on this server.

Is there any reason why assigning to a PKG var could be delayed 3 seconds.

2019-11-13T21:58:04.018631614Z 23(42) WARNING: <core> [core/action.c:1586]: run_actions(): alert - action [corefunc (16)] cfg [/etc/kamailio/kamailio.cfg:115] took too long [2579486 us]
$var(reply) = 0; // I wonder how this can endup blocking ?

2019-11-13T21:58:04.018748710Z 42(61) WARNING: <core> [core/action.c:1586]: run_actions(): alert - action [corefunc (16)] cfg [/etc/kamailio/kamailio.cfg:221] took too long [2222010 us]
} // Samething here, simply exiting a scope

2019-11-13T21:58:04.015675095Z 11(30) WARNING: <core> [core/action.c:1586]: run_actions(): alert - action [t_relay_cancel (24)] cfg [/etc/kamailio/kamailio.cfg:86] took too long [2588818 us]
} if (!t_relay_cancel()) { // Indeed a t_relay_cancel

2019-11-13T21:58:04.018647988Z 26(45) WARNING: <core> [core/action.c:1586]: run_actions(): alert - action [t_relay (24)] cfg [/etc/kamailio/kamailio.cfg:289] took too long [2739366 us]                    
t_relay(); // Indeed a t_relay

2019-11-13T21:58:04.018651243Z 15(34) WARNING: <core> [core/action.c:1586]: run_actions(): alert - action [sl_send_reply (26)] cfg [/etc/kamailio/kamailio.cfg:112] took too long [2587812 us]

sl_send_reply("100", "Trying"); // Indeed a sl_send_reply

Not sure which mailing list is best for this question

Regards

Julien