Library parralelism is set to 1 There is 2 drives (target session set to 4 for each one) There is 6 pools with 1 or 2 media for each one. savegrp parralelism is set to 0 for each savegrp
What do you think about that?
That's excatly what I was refering to Trekk.
Savegrp parallelism 0 means savegrp will run on parallelism which is equal to the sum of parallelism of each client in this group.
Ex. If there are four client in the group each with parallelism of 2 & savegrp parallelism is set to 0, it would mean savegrp parallelism is 4clientsx2 i.e. 8. Now Target session for the drive is 4. The group starts...picks up one media...starts pulling 8 streams...but since target session is 4, it can write first 4 streams to this media. Since the next 4 streams are available and waiting, it keeps flashing the waiting for...message. If it gets the second appendable media in this pool and second drive is free, it will pick up the second media and continue writing the next 4 streams there.
Why don't you test it by simplifying the scenario. When no groups are running, set savegrp parallelism of this group to 4 and start the group. Now since the group can write a max. of 4 streams at a time, it will start writing and would not wait for more media?
Hope I have been able to explain better this time.
This property of jukebox is called: "Auto Media Management" and have only two values available: True or False. This is a part of admin guide:
Auto Media Management gives the NetWorker server automatic control over media loaded in the storage device. When Auto Media Management is enabled during device configuration, the NetWorker server automatically: ◆ Labels the volume (recognizes EDM labels and does not overwrite them). ◆ Mounts the volume. ◆ Overwrites volumes it considers to be unlabeled. The NetWorker server considers a volume to be unlabeled when any of these conditions are true: ¿ It has no internal label. ¿ It is labeled with information other than a recognizable NetWorker label. ¿ It is labeled with a NetWorker label, but the density indicated on the internal label differs from that of the device where the volume is mounted. ¿ Recycles volumes eligible for reuse that are loaded into the device. ◆ Because the Auto Media Management feature can relabel a volume that has a different density, it is possible, inadvertently, to overwrite data that still has value. For this reason, be careful if NetWorker volumes are shared among devices with different densities. When the Auto Media Management feature is not enabled, the NetWorker server ignores unlabeled volumes and does not consider them for backup.
Networker is flooding the log with this kind of lines (and send me warning and alert emails) Why? There is a media for BACKUPPOOL in library, and he will succeed to backup the pool anyway. He is doing the same for all pool asking for a media despite the fact that there are one or many available and that he will always succeed to load one and write on it
Hi,
If I understand correctly, NW picks up one media from the pool and starts writing. However, it still gives these messages as if it is waiting for more writable media. If I am correct in understanding the scenario, then the problem is:
The groups have a larger parallelism then the Target sessions for the drives. Networker wants more media so that it can distribute more save streams across multiple media/drives.
Ex. A group has a parallelism of 10 & the drive has a Target session of 4. When the group starts, it will write first 4 save sterams on first drive & media. IT has 6 more streams available with it, therefore it looks for more writable media so that it can utilize the second drive with an appendable media to write the next 4 streams.
To resolve this,
1. Group parallelism can be made equal to Target session. 2. More media can be labelled for this group so that it writes on multiple media and does not give this error.
Please check that the server parallelism is also not too high. If Group parallelism is set to 0, then it is automatically considered equal to the sum of parallelism for the clients inside it.
Library parralelism is set to 1 There is 2 drives (target session set to 4 for each one) There is 6 pools with 1 or 2 media for each one. savegrp parralelism is set to 0 for each savegrp
amediratta
2 Intern
•
2K Posts
0
July 9th, 2007 10:00
Library parralelism is set to 1
There is 2 drives (target session set to 4 for each
one)
There is 6 pools with 1 or 2 media for each one.
savegrp parralelism is set to 0 for each savegrp
What do you think about that?
That's excatly what I was refering to Trekk.
Savegrp parallelism 0 means savegrp will run on parallelism which is equal to the sum of parallelism of each client in this group.
Ex. If there are four client in the group each with parallelism of 2 & savegrp parallelism is set to 0, it would mean savegrp parallelism is 4clientsx2 i.e. 8. Now Target session for the drive is 4. The group starts...picks up one media...starts pulling 8 streams...but since target session is 4, it can write first 4 streams to this media. Since the next 4 streams are available and waiting, it keeps flashing the waiting for...message. If it gets the second appendable media in this pool and second drive is free, it will pick up the second media and continue writing the next 4 streams there.
Why don't you test it by simplifying the scenario. When no groups are running, set savegrp parallelism of this group to 4 and start the group. Now since the group can write a max. of 4 streams at a time, it will start writing and would not wait for more media?
Hope I have been able to explain better this time.
benzino1
2 Intern
•
244 Posts
0
July 9th, 2007 00:00
Is the automedia management switched on on the library?
Trekk1
1 Rookie
•
48 Posts
0
July 9th, 2007 01:00
There are 4 choices :
Random
Sequential
Autoload
Loop
Random is checked
benzino1
2 Intern
•
244 Posts
0
July 9th, 2007 02:00
This property of jukebox is called: "Auto Media Management" and have only two values available: True or False. This is a part of admin guide:
Auto Media Management gives the NetWorker server automatic control over media loaded
in the storage device. When Auto Media Management is enabled during device
configuration, the NetWorker server automatically:
◆ Labels the volume (recognizes EDM labels and does not overwrite them).
◆ Mounts the volume.
◆ Overwrites volumes it considers to be unlabeled.
The NetWorker server considers a volume to be unlabeled when any of these conditions
are true:
¿ It has no internal label.
¿ It is labeled with information other than a recognizable NetWorker label.
¿ It is labeled with a NetWorker label, but the density indicated on the internal label
differs from that of the device where the volume is mounted.
¿ Recycles volumes eligible for reuse that are loaded into the device.
◆ Because the Auto Media Management feature can relabel a volume that has a different
density, it is possible, inadvertently, to overwrite data that still has value. For this reason,
be careful if NetWorker volumes are shared among devices with different densities.
When the Auto Media Management feature is not enabled, the NetWorker server ignores
unlabeled volumes and does not consider them for backup.
benzino1
2 Intern
•
244 Posts
1
July 9th, 2007 05:00
Trekk1
1 Rookie
•
48 Posts
0
July 9th, 2007 05:00
Should I enable it?
Trekk1
1 Rookie
•
48 Posts
0
July 9th, 2007 06:00
I will tell you tomorrow if Networker is happier during backup task
amediratta
2 Intern
•
2K Posts
1
July 9th, 2007 08:00
Networker is flooding the log with this kind of lines
(and send me warning and alert emails)
Why?
There is a media for BACKUPPOOL in library, and he
will succeed to backup the pool anyway. He is doing
the same for all pool asking for a media despite the
fact that there are one or many available and that he
will always succeed to load one and write on it
Hi,
If I understand correctly, NW picks up one media from the pool and starts writing. However, it still gives these messages as if it is waiting for more writable media. If I am correct in understanding the scenario, then the problem is:
The groups have a larger parallelism then the Target sessions for the drives. Networker wants more media so that it can distribute more save streams across multiple media/drives.
Ex. A group has a parallelism of 10 & the drive has a Target session of 4. When the group starts, it will write first 4 save sterams on first drive & media. IT has 6 more streams available with it, therefore it looks for more writable media so that it can utilize the second drive with an appendable media to write the next 4 streams.
To resolve this,
1. Group parallelism can be made equal to Target session.
2. More media can be labelled for this group so that it writes on multiple media and does not give this error.
Please check that the server parallelism is also not too high. If Group parallelism is set to 0, then it is automatically considered equal to the sum of parallelism for the clients inside it.
Trekk1
1 Rookie
•
48 Posts
0
July 9th, 2007 09:00
Library parralelism is set to 1
There is 2 drives (target session set to 4 for each one)
There is 6 pools with 1 or 2 media for each one.
savegrp parralelism is set to 0 for each savegrp
What do you think about that?