ZoneMinder and NAS
Posted: Fri Aug 21, 2009 8:19 am
Using 1.24.2 on Fedora Core 7, CPU is Intel Core 2 Duo, 1 GIG of RAM
To be able to separate the storage from the processing machine, I bought a Synology DS209 NAS box. I connect this box using a GBit Ethernet switch to the ZoneMinder server using NFS, so that i mount
/NAS_IP/volulme/exported_dir -> /home/www/events
NFS connection is stable with a 12MByte/s burst read/write speed. However, during operations, Zoneminder drops a lot of messages into /var/log/messages about "Incomplete frame read" and "Incomplete frame write". Yet the NAS BOX contains all the images in sequence 001, 002, etc. They are readable, displayable.
Now once or twice a day up to 10 STRANGE events appear in the event listing. Their duration is extermely long (several millions of seconds), yet contain only up to 100 frames. Such events can't be viewed, they display only black images and ZM gets completely STUCK (no web response). The only solution is to restart the ZM service.
I don't yet know how do those events appear in the DB, but they only appear if the events folders is mapped to an NFS disk. The messages about "Incomplete frame read/write" also appear ONLY if I am using a NFS mapped directory to store the images. Ring buffers and SHMEM is set correctly, no messages in the logs about them.
If I let ZM write images to local disks, ZM is pretty stable.
Please advise, what to test and how to configure ZM to work through NFS as good as with local disks???
To be able to separate the storage from the processing machine, I bought a Synology DS209 NAS box. I connect this box using a GBit Ethernet switch to the ZoneMinder server using NFS, so that i mount
/NAS_IP/volulme/exported_dir -> /home/www/events
NFS connection is stable with a 12MByte/s burst read/write speed. However, during operations, Zoneminder drops a lot of messages into /var/log/messages about "Incomplete frame read" and "Incomplete frame write". Yet the NAS BOX contains all the images in sequence 001, 002, etc. They are readable, displayable.
Now once or twice a day up to 10 STRANGE events appear in the event listing. Their duration is extermely long (several millions of seconds), yet contain only up to 100 frames. Such events can't be viewed, they display only black images and ZM gets completely STUCK (no web response). The only solution is to restart the ZM service.
I don't yet know how do those events appear in the DB, but they only appear if the events folders is mapped to an NFS disk. The messages about "Incomplete frame read/write" also appear ONLY if I am using a NFS mapped directory to store the images. Ring buffers and SHMEM is set correctly, no messages in the logs about them.
If I let ZM write images to local disks, ZM is pretty stable.
Please advise, what to test and how to configure ZM to work through NFS as good as with local disks???