In this Post I will show you how to remove one or more Cages completely or some disks from a 3PAR Storage system. This process involve some risks but it can be done completely Online and without disruption.
Let's first see the requirements for this operation.
Some requirements for this process:
- Backup all the data on the Storage;
- If you have RemoteCopy in place with RCIP or RCFC, you should stop the replication;
- This operation takes some time (depending the size ans speed of the disks);
- Do it after hours, this change it's very disk IOPS consuming;
- You must have free space on the remaining cages to accommodate all of you data;
- It's not mandatory, but having a HPE support contract it's kind a handy;
- Change to maintenance mode in Service Processor;
- Migrate/move all the data from the disk cages you want to remove. This can be done with the following:
- If you have one or more CPG with all the same type of disks (the standard procedure), you must create a new CPG with disk filtering enable so you can use all the disks that are not on the cage to be removed;
- Use Tune Virtual Volume (VV) to move the Data to this new CPG. You need Dynamic Optimization license to tune VVs or Create new Volumes on the new CPG and if you use VMware/Hyper-V, you can Storage vMotion to the new disks;
- When using Adaptive Optimization (AO), change the AO-configuration to remove CPG that contains disks of cage to remove;
- After all the migrations you can delete the old Volumes and CPGs;
- Verify that the disks on the cage(s) to be removed have only spare chunklets (you can type showpd -c)
After that, we can begin the process of Cage/Disk removal.
As I mentioned before, be aware that this involves some risk and you must have some plan B like Backups outside the Storage. So, start with a SSH to the storage system.
- View all the disks on the Storage
- showpd -c
- Check and Note the IDs of the disks on the cage(s) you want to remove
- Remove spare chunklets from each disk to remove
- removespare -f n:a (n for physical disk ID)
- Check if disk has no user/spare data
- showpd -c
- Every disk in the cage must have 0GB used data (Column: Used | OK). If so, you are lucky! If not, you can do the following:
- showpdch n (n for PD-ID)
- You will probably see data like: admin*, .srdata*, pdsld* or log* system meta-data is stored on disks you want to remove. admin and .srdata are system-VVs.
- You can check which disks these VVs are using by running showvvpd admin and showvvpd .srdata.
- If you see pdsld* or log*:
stopsr -f
stoppdata
removeld -rmsys logX.Y (0, 1 or 2)
removeld -rmsys pdsldX.Y (0, 1 or 2)
- If you see admin*, .srdata*:
movepdtospare -f -vacate -nowait n (n for physical disk ID)
movepdtospare -nowait -f -perm -devtype -ovrd (n for physical disk ID)
- If you can't remove the data with the above commands, contact HPE Support to migrate data to other disk (types) in the system. Especially the .srdata and admin can be hard to move.
- Dismiss the physical disk
- dismisspd n (n for PD-ID)
- This command removes disks from configuration. After this command, disks must be stated as "new".
- If an error says that disk is in use, go to the step above and repeat the process
- Is you see errors like these:
- Error : Pd id XX is referenced by chunklet Y:Z
- Run the following command for every chunklet
- cli% movech -perm -ovrd -f Y:Z
- Turn the disks Off (Optional)
- controlmag offloop -f cageX n (n for physical disk ID and X for the cage number)
- Stop here if you only want to remove disks. After the dismiss you can pull out the disk (don't forget that the 3PAR cages must have equal number of same type of disks or a difference of two)
- Remove Cage from configuration
- servicecage remove -f cageX
- If there is an error, you can put Offline the connection in SSMC/3PAR Management Console or physically remove the SAS cables and then try again the above command
- Remove cables from cage and turn off power supplies
- Reconfigure the cable connection without the cage you just removed. Check HPE 3PAR Cabling Guide.
- Post Configurations
- admithw
- startsr -f
- checkhealth (to re-calculate spare-chunklets)
I hope it helped and if you have any questions let me know.
See you next time!
Thank you very much!
ReplyDeleteWonderfull article!
Thanks a lot for the article! I've played a little bit with 3PAR 7400c initially equipped with HDD's and upgraded with SSD's later. I've been able to move log LD's but not "admin" or ".srdata". But, it is interesting, what will happen if I will move "admin" LD chunklets to SSD's by issuing "movech" manually for each chunklet? Will the storage boot ok after that and how to start chunklets redistribution in future?
ReplyDeleteHello,
DeleteAs far as i know it is necessary to move all .admin and .srdata data to the new disks with movepdtospare command.
The redistribution is done automatically by the system and you can run tunesys command to analyze the entire system and automatically corrects space usage imbalances.
What would be the procedure to remove a failed disk if you dont want to replace it? Can in it be just removed after servicemag is completed?
ReplyDeleteHello, if you want to remove only the disk, you can use the command "dismisspd n" (n for PD-ID).
DeleteThis command removes disks from configuration. Don't forget that the 3PAR cages must have equal number of same type of disks or a difference of two. After that you can tunesys to balance the data across all the disks according to the defined RAID.