> If it is possible to run the clean while the other drives are in use, how do I do so? ANS=yes...
Hardware cleaning (i.e. jukebox enabled auto clean) must be disabled while NetWorker is controlling the jukebox. The reason is that very few jukeboxes (a.k.a. autochangers, tape libraries) are intelligent enough to send a signal to the application to "pause" so that the jukebox can perform a cleaning. Without this ability, the drives go offline, and NetWorker sees this as an i/o error. After a number of consecutive errors on the drive (default is 20), NetWorker will set the drive status to disable to prevent further usage.
If you would like NetWorker to perform auto cleaning, you should disable hardware cleaning on the jukebox, and load a cleaning tape into one of the data slots (usually the last slot). You then set in the NetWorker jukebox properties:
set auto clean to if you want it to auto clean periodically (Default is 6 months, this is set in device properties)
set the cleaning slot(s)
set the default number of usage for a unused cleaning tape.
In some jukeboxes, there can be a dedicated slot used to hold a cleaning tape. Do not use this slot, because this dedicated slot is not part of the data slots used to hold tapes. Also, do not enable hardware cleaning. Put the cleaning tape in one of the data slots that you assign.
If you use NetWorker Management Console (NMC) to configure the jukebox, the cleaning slot will automatically set as the last data slot.
To set or reset the number of usages, please run:
nsrjb -S (cleaning slot) -U (number of usage remaining)
If the drive requires a cleaning, and there is a cleaning slot loaded with a cleaning tape, then :
If the drive is idle, then NetWorker will wait until it selects the drive for backup/recover operation. It will then clean the drive before actually loading a data tape
If the drive is busy, then NetWorker will wait until the backup/recover operation is completed, and then perform the cleaning.
To manually force a cleaning, you can do one of of the following:
In the NetWorker tape drive device properties, set "Cleaning Required" to Yes. The cleaning will be scheduled based on whether is busy or not (see above).
If the drive is idle, use NMC to move the cleaning tape in the cleaning slot to the drive.
If the drive is idle, physically insert the cleaning tape into the drive. Physically remove once finished.
Note:
Contact your supplier to determine how many times the cleaning tape can be used. NetWorker currently assumes just 5. If this is not set correctly, then NetWorker would stop using that cleaning tape prematurely, which may be costly.
NetWorker requires cleaning tape cartridges to have a bar code label.
If you are using a virtual tape library (VTL), please turn off auto cleaning, and do not set the cleaning slot. I know VTL's are suppose to emulate physical tape libraries as closely as possible, however there is absolutely no need to clean a virtual tape drive.
A tape cleaning is initiated by inserting the cleaning tape into the drive. Of course you cannot mount a cleaning tape.
It will automatically be ejected after some seconds/minutes. There is no necessity to take the library offline. NW will not do that but of course the library firmware could do that (which is somehow stupid).
As Wallace mentioned, NW could do the job automatically
- In very old times, this was done based on a timer which lead to cleaning cycles although a drive might not have been used at all.
- since NW 7? he can act on appropriate 'Tape Alerts' issued by the drive. If cleaning is enabled, he can react upon request.
So there is no real need to run a manual cleaning in advance.
Thanks for the replies. So to summarize, there's no reason why Networker cannot clean one drive while backing up to the other drives, and to launch the clean I just load (without mount) the cleaning tape into the drive, assuming that the "require cleaning" flag is set to yes.
I've heard that having Networker manage the cleaning is unreliable. Is that still the case?
I turned on auto-clean, expecting Networker to wait for the drive to be idle before attempting to clean it. Instead, it gives me a bunch of errors that it's "Automatically terminating operation 'OP_CLEAN'... Cannot allocate the 1 required device(s)."
Any ideas? I've turned auto-clean back off so I don't get continuous errors. The drive's not going to be free for several hours.
Thanks for the reply, but I thought the whole point of auto-clean was that Networker would wait for the drive to be idle before attempting to clean it?
If the library is idle then I can just run the cleaning cycle from the tape library GUI, as I usually do. I'm trying to see if I can run it while the library is doing backups using the other tape drives, preferably automatically so I don't have to wait until the drive is temporarily idle (i.e. it reached the end of the tape and the next tape is loaded into a different drive). My backup jobs are a week long so I'm risking an error and dropped backup by not cleaning the tape drive.
"Cannot allocate the 1 required device(s)." implies that the NetWorker drive that you are trying to use is still busy. This can be because there is some read or write in progress, or it can also mean that the device is reserved for some pending work.
If the cleaning is not urgent, then I suggest that you wait until there is no tape drive activity, then perform your cleaning.
However, if you need to perform the cleaning, then you can still manually clean the drive by physically insert the cleaning tape into the drive and then remove it once done. IF the drive is not being used at the moment, then you can disable the jukebox in NetWorker and then move the cleaning tape in and out of the tape drive by either:
use the jukebox control to move the cleaning tape into the drive
on the host that controls the jukebox (e.g. NetWorker server) , use sjimm command
open the jukebox and manually load and unload the cleaning cartridge
Once the cleaning is done, re-enable the NetWorker jukebox.
> I'm trying to see if I can run it while the library is doing backups using the other tape drives, preferably automatically so I don't have to wait until the drive is temporarily idle (i.e. it reached the end of the tape and the next tape is loaded into a different drive). My backup jobs are a week long so I'm risking an error and dropped backup by not cleaning the tape drive.
If the tape drives are in use, then NetWorker has to wait till they are not in use before NetWorker can clean the drives. Not in use means not allocated, and not reading or writing.
An example of tape drive being allocated is during a cloning operation, where two drives are needed. In this scenario, the drive used for reading maybe be loaded with the volume for reading, but there is no drive available for writing. So the drive that is loaded for reading appears idle, but in fact it is still allocated because of the pending clone job.
If you believe that the drive that you want to clean is really idle and not in use, then you can reset the drive using the following NetWorker command, which will reset the corresponding nsrmmd process that controls the tape drive:
nsrmm -HHv -f (NetWorker device name)
You then may want to perform a quick inventory, so that NetWorker is up to date with the current volume list:
nsrjb -IIEv
If you are unsure, or require additional assistance, then please open a live chat, or a new service request so that a technical support person can look at and assess your NetWorker jukebox environment.
I'm not sure you're understanding my question. I realize the drive has to be idle before Networker can clean it. What I want to know is whether, when set to auto-clean, Networker is expected to wait for the drive to be idle before attempting the clean. Or does it just try anyway, regardless of whether the drive is idle, and thus produce errors every few minutes? Is that normal behavior?
If that is normal, then what's the point of auto-clean? Is it really designed to just keep trying (and erroring out) until the drive is idle?
When set to auto-clean, NetWorker is expected to wait for the drive to be idle before attempting the clean. Or does it just try anyway, regardless of whether the drive is idle, and thus produce errors every few minutes? Is that normal behavior?
Both answers=Yes
If that is normal, then what's the point of auto-clean? Is it really designed to just keep trying (and erroring out) until the drive is idle?
As stated earlier, the jukebox auto clean is used to periodically clean the tape drives. The time interval between is set at the NetWorker device properties. However, automatic cleaning can also be triggered after it receives a “device needs cleaning” notification.
Environment variable MGD_DEV_ALLOC_RETRIES can be used to increase number of retries for device allocation. By default NetWorker (nsrmmgd) tries device allocation once every 2 seconds, and the default value for device allocation is 30 retries, which means 60 seconds or one minute timeout value.
Increase default number of retries from 30 to 300 (which means increasing timeout value from 1 minute to 10 minutes):
Define the following environment value on NetWorker server.
MGD_DEV_ALLOC_RETRIES=300
export MGD_DEV_ALLOC_RETRIES
Restart NetWorker services to get new setting applied.
Please give this a try. If this does not work, then engineering may not have implemented this in NetWorker 9. If so, please open a new service request to discuss this matter.
wlee
263 Posts
0
April 6th, 2017 18:00
> If it is possible to run the clean while the other drives are in use, how do I do so? ANS=yes...
Hardware cleaning (i.e. jukebox enabled auto clean) must be disabled while NetWorker is controlling the jukebox. The reason is that very few jukeboxes (a.k.a. autochangers, tape libraries) are intelligent enough to send a signal to the application to "pause" so that the jukebox can perform a cleaning. Without this ability, the drives go offline, and NetWorker sees this as an i/o error. After a number of consecutive errors on the drive (default is 20), NetWorker will set the drive status to disable to prevent further usage.
If you would like NetWorker to perform auto cleaning, you should disable hardware cleaning on the jukebox, and load a cleaning tape into one of the data slots (usually the last slot). You then set in the NetWorker jukebox properties:
In some jukeboxes, there can be a dedicated slot used to hold a cleaning tape. Do not use this slot, because this dedicated slot is not part of the data slots used to hold tapes. Also, do not enable hardware cleaning. Put the cleaning tape in one of the data slots that you assign.
If you use NetWorker Management Console (NMC) to configure the jukebox, the cleaning slot will automatically set as the last data slot.
To set or reset the number of usages, please run:
If the drive requires a cleaning, and there is a cleaning slot loaded with a cleaning tape, then :
To manually force a cleaning, you can do one of of the following:
Note:
bingo.1
2.4K Posts
0
April 6th, 2017 21:00
A tape cleaning is initiated by inserting the cleaning tape into the drive. Of course you cannot mount a cleaning tape.
It will automatically be ejected after some seconds/minutes. There is no necessity to take the library offline. NW will not do that but of course the library firmware could do that (which is somehow stupid).
As Wallace mentioned, NW could do the job automatically
- In very old times, this was done based on a timer which lead to cleaning cycles although a drive might not have been used at all.
- since NW 7? he can act on appropriate 'Tape Alerts' issued by the drive. If cleaning is enabled, he can react upon request.
So there is no real need to run a manual cleaning in advance.
brastedd
2 Intern
•
336 Posts
0
April 7th, 2017 04:00
Thanks for the replies. So to summarize, there's no reason why Networker cannot clean one drive while backing up to the other drives, and to launch the clean I just load (without mount) the cleaning tape into the drive, assuming that the "require cleaning" flag is set to yes.
I've heard that having Networker manage the cleaning is unreliable. Is that still the case?
brastedd
2 Intern
•
336 Posts
0
April 7th, 2017 06:00
I turned on auto-clean, expecting Networker to wait for the drive to be idle before attempting to clean it. Instead, it gives me a bunch of errors that it's "Automatically terminating operation 'OP_CLEAN'... Cannot allocate the 1 required device(s)."
Any ideas? I've turned auto-clean back off so I don't get continuous errors. The drive's not going to be free for several hours.
brastedd
2 Intern
•
336 Posts
0
April 7th, 2017 07:00
Thanks for the reply, but I thought the whole point of auto-clean was that Networker would wait for the drive to be idle before attempting to clean it?
If the library is idle then I can just run the cleaning cycle from the tape library GUI, as I usually do. I'm trying to see if I can run it while the library is doing backups using the other tape drives, preferably automatically so I don't have to wait until the drive is temporarily idle (i.e. it reached the end of the tape and the next tape is loaded into a different drive). My backup jobs are a week long so I'm risking an error and dropped backup by not cleaning the tape drive.
Thanks.
wlee
263 Posts
0
April 7th, 2017 07:00
"Cannot allocate the 1 required device(s)." implies that the NetWorker drive that you are trying to use is still busy. This can be because there is some read or write in progress, or it can also mean that the device is reserved for some pending work.
If the cleaning is not urgent, then I suggest that you wait until there is no tape drive activity, then perform your cleaning.
However, if you need to perform the cleaning, then you can still manually clean the drive by physically insert the cleaning tape into the drive and then remove it once done. IF the drive is not being used at the moment, then you can disable the jukebox in NetWorker and then move the cleaning tape in and out of the tape drive by either:
Once the cleaning is done, re-enable the NetWorker jukebox.
wlee
263 Posts
0
April 7th, 2017 08:00
> I'm trying to see if I can run it while the library is doing backups using the other tape drives, preferably automatically so I don't have to wait until the drive is temporarily idle (i.e. it reached the end of the tape and the next tape is loaded into a different drive). My backup jobs are a week long so I'm risking an error and dropped backup by not cleaning the tape drive.
If the tape drives are in use, then NetWorker has to wait till they are not in use before NetWorker can clean the drives. Not in use means not allocated, and not reading or writing.
An example of tape drive being allocated is during a cloning operation, where two drives are needed. In this scenario, the drive used for reading maybe be loaded with the volume for reading, but there is no drive available for writing. So the drive that is loaded for reading appears idle, but in fact it is still allocated because of the pending clone job.
If you believe that the drive that you want to clean is really idle and not in use, then you can reset the drive using the following NetWorker command, which will reset the corresponding nsrmmd process that controls the tape drive:
You then may want to perform a quick inventory, so that NetWorker is up to date with the current volume list:
If you are unsure, or require additional assistance, then please open a live chat, or a new service request so that a technical support person can look at and assess your NetWorker jukebox environment.
brastedd
2 Intern
•
336 Posts
0
April 7th, 2017 10:00
I'm not sure you're understanding my question. I realize the drive has to be idle before Networker can clean it. What I want to know is whether, when set to auto-clean, Networker is expected to wait for the drive to be idle before attempting the clean. Or does it just try anyway, regardless of whether the drive is idle, and thus produce errors every few minutes? Is that normal behavior?
If that is normal, then what's the point of auto-clean? Is it really designed to just keep trying (and erroring out) until the drive is idle?
wlee
263 Posts
0
April 7th, 2017 12:00
Both answers=Yes
As stated earlier, the jukebox auto clean is used to periodically clean the tape drives. The time interval between is set at the NetWorker device properties. However, automatic cleaning can also be triggered after it receives a “device needs cleaning” notification.
Perhaps the errors are coming too frequently. As discussed in "Backup hangs with media waiting message after tape cleaning operation" (https://support.emc.com/kb/488937) for NetWorker 8.2.x:
Environment variable MGD_DEV_ALLOC_RETRIES can be used to increase number of retries for device allocation. By default NetWorker (nsrmmgd) tries device allocation once every 2 seconds, and the default value for device allocation is 30 retries, which means 60 seconds or one minute timeout value.
Increase default number of retries from 30 to 300 (which means increasing timeout value from 1 minute to 10 minutes):
Please give this a try. If this does not work, then engineering may not have implemented this in NetWorker 9. If so, please open a new service request to discuss this matter.