[FortiCron] memory leak in Forticron process ?

[beufa]

Hi all, We have seen that some problems in Forticron process, which appears to have memory leak in 4.0 MR3 (all patchs). Memory consumption increases for this process around 50% each 15days ... Is anyone already seen this and report the problem ? Thanks for your feedback !

[KKedda]

We' ve had the same problem. Were running on v 4.0 MR3 P2. When running diag sys top, forticron was using 33.5 % memory after 200+ days uptime. With FortiNet on the phone, restarted the forticron process and the forticron process memory usage went down to 0.5%. Just upgraded to v 4.0 MR3 P9. After 8 hours, memory utilization is steady at 55%. Previously ran at 85%. I' m putting it back into production tonight. I hope I chose the right version of FortiOS!

[charles_demers_tremb]

What' s the command to restart the forticron process ? Is there any impacts ?

[beufa]

no known impacts seen. Use diag sys top to find the forticron process ID and after do a diag sys kill [9|11] PID. 4.0MR3 P9 seems to be more stable, but in one week usage, we already see +2% of memory usage., due to forticron increase. So problem is not fixed. I think I will open a ticket on support.

[charles_demers_tremb]

Thx Killing forticron we managed to " save" 15 % of memory on a very busy cluster We hope 4.3.9 will be more memory friendly as we really have trouble running some services (vpnssl...ips...)

[beufa]

There is no fix in 4.0 MR3 Patch 9 of this problem. I will open a ticket this afternoon, but forticron' s memory problem is not fixed ... Still +30% of memory each 15 days .....

[charles_demers_tremb]

Maybe patch 10 fix this ?

[beufa]

Official reply by Fortinet : " A bug report was opened for this problem in July. Its status is now resolved for 5.0 release which is planed to be released in Q4." " In the bug notes it is stated that the issue is resolved already for 5.0 but there is no plan to release a fix for 4 MR3." But no chances I will put FortiOS v5.0 in production as soon as it' s released.... Perhaps it will be corrected in later MR3 Patches ...