cancel
Showing results for 
Search instead for 
Did you mean: 

aplicc

tanislavm
Level 6
Hi, If I have 2 groups running on an vcs node,how could I find out what processes belong to group A and what processes belong to group B?Maybe if I look at user and group of every process,and this should to meet the user and group when I issue ls -l /dir/application? I like to control every step in recovery of an volume with 2 plexes on stale. Based on size and flag I will select the plex I wish to recover first.right?Then with vxrecover I start the volume with this plex.Then I ask to the user who uses this volume to check if all the files are in place.If yes,then I recover the second plex.During this recovering the user could work safely.right? thanks so much.
1 ACCEPTED SOLUTION

Accepted Solutions

Gaurav_S
Moderator
Moderator
   VIP    Certified

You say that in this case i should to offline also the 2nd plex(with good data),so now the 2 plexes are offline.right?then vxrecover plex_with_gooddata?right?

>> No, that is not what I am saying. If the 2nd is plex is running, why would you want to take offline ? If second plex fails as well, then you should recover 2nd plex first because it had updated data.You will need to use vxplex, vxmend, vxvol depending on what state plex is in.

But what about the case when i recover first plex,and user says that files are missing,then i recover the 2nd plex and the same from this user.should i use the restoration from backup?

>> If by accident you start 1st plex first & user says files are missing, once you trigger vxrecover or attach second plex in mirror, contents of 1st plex will copy on 2nd so you will need to initiate restore from backup. But if you are sure that 2nd plex has good data, then you can offline 1st plex, recover 2nd plex & start volume using 2nd plex, once started & user verified, attach 1st plex as mirror to 2nd plex.

 

G

View solution in original post

3 REPLIES 3

Gaurav_S
Moderator
Moderator
   VIP    Certified

Hi,

If I have 2 groups running on an vcs node,how could I find out what processes belong to group A and what processes belong to group B?Maybe if I look at user and group of every process,and this should to meet the user and group when I issue ls -l /dir/application?

>>> This needs manual mapping, there will be no such flag which will indicate whether a process belongs to group A or B. If the application is started with specific user credientials then yes you can but this also may be confusing. For e.g if both service group contains Oracle DB, both may use Oracle user name to start so you won't be able to identify with username, rather in this example of DBs, instance name can be an identifier..

I like to control every step in recovery of an volume with 2 plexes on stale.
Based on size and flag I will select the plex I wish to recover first.right?Then with vxrecover I start the volume with this plex.Then I ask to the user who uses this volume to check if all the files are in place.If yes,then I recover the second plex.During this recovering the user could work safely.right?

>> Not based on size, but should be based on structure of volume. If its a mirrored volume & if both the plexes were active, you can choose to recover any plex first. If it was a mirrored volume & 1 plex was already down (that means it doesn't have all the updates) & then you hit second plex down as well (which had updated data), then you would want to recover later plex as it had updated data.

Then I ask to the user who uses this volume to check if all the files are in place.If yes,then I recover the second plex.During this recovering the user could work safely.right?

>> Correct

 

G

tanislavm
Level 6

Hi,

You state:

If it was a mirrored volume & 1 plex was already down (that means it doesn't have all the updates) & then you hit second plex down as well (which had updated data), then you would want to recover later plex as it had updated data.

 

 

You say that in this case i should to offline also the 2nd plex(with good data),so now the 2 plexes are offline.right?then vxrecover plex_with_gooddata?right?

 

then ask the user to verify the files.

 

But what about the case when i recover first plex,and user says that files are missing,then i recover the 2nd plex and the same from this user.should i use the restoration from backup?

 

 

 

thanks so much.

 

 

Gaurav_S
Moderator
Moderator
   VIP    Certified

You say that in this case i should to offline also the 2nd plex(with good data),so now the 2 plexes are offline.right?then vxrecover plex_with_gooddata?right?

>> No, that is not what I am saying. If the 2nd is plex is running, why would you want to take offline ? If second plex fails as well, then you should recover 2nd plex first because it had updated data.You will need to use vxplex, vxmend, vxvol depending on what state plex is in.

But what about the case when i recover first plex,and user says that files are missing,then i recover the 2nd plex and the same from this user.should i use the restoration from backup?

>> If by accident you start 1st plex first & user says files are missing, once you trigger vxrecover or attach second plex in mirror, contents of 1st plex will copy on 2nd so you will need to initiate restore from backup. But if you are sure that 2nd plex has good data, then you can offline 1st plex, recover 2nd plex & start volume using 2nd plex, once started & user verified, attach 1st plex as mirror to 2nd plex.

 

G