[clamav-users] how to narrow down the signature database?

2015-11-20 Thread Bond Masuda
my question is in the subject line, but my goal is to reduce the memory footprint of clamd. it current takes over 350MB and that's a bit too much. so, i'm wondering how I can narrow down the signature database to reduce this memory footprint. specifically, i only care about malware that is relevan

Re: [clamav-users] clamdscan troubleshooting

2015-11-21 Thread Bond Masuda
Daniel, You might want to look at these two SELinux booleans: antivirus_can_scan_system antivirus_use_jit You can use 'getsebool': $ getsebool antivirus_can_scan_system antivirus_can_scan_system --> on And you can use 'setsebool' to toggle the boolean setting. Additionally, see man page for c

Re: [clamav-users] ScanOnAccess issue when clamd launched from systemd

2016-05-04 Thread Bond Masuda
Is SELinux enabled on that system? If so, I would look to see if SELinux is blocking. On 05/04/2016 09:29 AM, Mikko Caldara wrote: Hello, I'm trying to configure OnAccess scanning on the whole drive, as read in this post blog: http://blog.clamav.net/2016/03/configuring-on-access-scanning-in-c

Re: [clamav-users] ScanOnAccess issue when clamd launched from systemd

2016-05-06 Thread Bond Masuda
Mikko, I suspected as much based on your description. Use 'ps -efZ | grep clamd' to find out how the clamd process is running. Check it when you run manually, and check it again when started by systemd. I suspect you will find a difference. Once you know the difference, then you can adjust the SE

Re: [clamav-users] ClamAV ScanOnAccess not scanning RHEL7

2017-03-31 Thread Bond Masuda
Remi, This is just a guess in the dark, but I've often encountered issues like that (service doesn't work started from systemd, yet works started from commandline) as a result of some SELinux AVC. Do you have SELinux enabled? And if so, does your test yield the desired result if you disable/m

Re: [clamav-users] ClamAV UnOfficial Database

2017-05-04 Thread Bond Masuda
I have to agree on the memory requirements for clamd being high. I wonder if it would make sense to store the data set of signatures in a tiered hierarchy, with some of the less used data being on disk until there's a hit? i don't know enough about the internals and scanning algorithm used in c