Troubleshooting
Problem
Procedure that allows you to collect both a Full IBM i Main Storage Dump (MSD), and a Subset MSD. This allows the subset to be later offloaded and transferred to IBM for analysis, while maintaining the ability to go back to the Full dump, if the need arises.
Resolving The Problem
What follows is a procedure that allows you to collect both a Full IBM i Main Storage Dump (MSD), and a Subset MSD. This allows the subset to be later offloaded and transferred to IBM for analysis, while maintaining the ability to go back to the Full dump, if the need arises.
The following screenshots are from a 7.2 system.
To check the current Main Storage Dump Copy Options on a system that is up and accessible, issue the STRSST command, sign in to SST, take 1. Start a service tool, 6. Main storage dump manager, and 3. Work with copy options.
A forced MSD will terminate the partition, and IPL to the MSD manager.
Press Enter to go to the MSD Manager menu
If you were not able to check the copy options at SST (i.e. system was hung, or SST was inaccessible prior to dumping), you can take option 3 here, and check the Subset value.
If the dump has already completed copying, you may have seen 'Copy completed normally.' status at the bottom of the screen.
If not, take the option 1. Work with current main storage dump (MSD), and then hit F11=Copy status.
You can watch the MSD status, including the Storage Management Recovery steps. Be aware that 1-3%, and 99%, cover special functions that can make them take longer than the other percentages. Do not estimate dump copy time from those percentages.
Once the copy completes, exit back to the MSD Manager menu. Do not exit past that, as you could start the IPL back up before we collect the other dump type.
Take option 3. Work with copy options. If the 'Subset if possible' value is set to '1=Yes', as is default at 7.2, change it to '3=No' and hit Enter. If the 'Subset if possible' value is set to '3=No', as is default prior to IBM i 7.2, change it to '1=Yes' and hit Enter.
This will take you back to the MSD Manager menu, and you should see a 'Copy options accepted.' status message.
Take option 1. Work with current main storage dump (MSD), followed by option 3. Copy to ASP
If you changed the 'Subset if possible' value to 3, then put the description 'Forced MSD - Full Copy'.
If you changed the 'Subset if possible' value to 1, then put the description 'Forced MSD - Subset Copy'.
Hit Enter
Since Storage Management Recovery has already completed, then the copy time will only be the time to actually write the dump to disk.
Once the copy completes, you will be put back to the Work with Current Main Storage Dump menu. Hit F12 to exit to the Main Storage Dump Manager menu. Take option 2. Work with copies of main storage dumps to confirm you have two copies of the dump.
*** Do NOT issue a 5 to display either of the dumps, as that will start an indexing process of the compressed dumps, which can take an extended period of time. ***
Once that is confirmed, F12 back to the Main Storage Manager, and then take option 3. Work with copy options. Change the 'Subset if possible' value back to its original value, and hit Enter.
This will take you back to the MSD Manager menu, and you should see a 'Copy options accepted.' status message.
You can now hit F3 twice to start the IPL back up.
Post-IPL, the recommendation is to submit the Subset copy of the dump to IBM first.
Other useful MSD topics
Copy MSD directly to and from IFS file (IBM i 7.2 and later) - http://www-01.ibm.com/support/docview.wss?uid=nas8N1020052
Customer Instructions for Copying a MSD from ASP (MSD Manager) to a Virtual Tape Image (prior to IBM i 7.2) - http://www-01.ibm.com/support/docview.wss?uid=nas8N1018687
Running Advanced Analysis macro SERVICEDOCS against an MSD - MSD Handling for Very Large Systems, See Option 3 to collect servicedocs against the MSD and send it to IBM https://www.ibm.com/support/pages/msd-handling-very-large-systems
Background for using this procedure
Since IBM i v5r2m0, system crashes had the potential to create a subset dump, if the SRC matched known failure types. In those cases, the system could determine which data was needed, or not needed, to diagnose the failure. However, forced MSDs had broadly been left in the "Can't know why you forced a dump." category. So, we preserved a full copy of partition memory. Over the years this approach has become more challenging as partition memory sizes have gone to the multi-Terabyte ranges. Despite compression of the dumps, the collection and transfer of these dumps can require significant time and resources.
At IBM i 7.2, and PTF'd back to IBM i 6.1.1 and 7.1, we have the function to create a subset dump for forced dumps. This can constitute a greater than 90% space savings in some cases, along with reduced collection and transfer times. At IBM i 6.1.1 and 7.1, the function is disabled by default. At IBM i 7.2, it is enabled by default. The concern is that, with the wide variety of reasons for forcing a dump, that subsetting has the potential to omit needed diagnostic data. As stated above, we needed a procedure that allows you to collect both a Full IBM i Main Storage Dump (MSD), and a Subset MSD, maintaining the ability to go back to the Full dump, if the need arises.
The following screenshots are from a 7.2 system.
To check the current Main Storage Dump Copy Options on a system that is up and accessible, issue the STRSST command, sign in to SST, take 1. Start a service tool, 6. Main storage dump manager, and 3. Work with copy options.
A forced MSD will terminate the partition, and IPL to the MSD manager.
Press Enter to go to the MSD Manager menu
If you were not able to check the copy options at SST (i.e. system was hung, or SST was inaccessible prior to dumping), you can take option 3 here, and check the Subset value.
If the dump has already completed copying, you may have seen 'Copy completed normally.' status at the bottom of the screen.
If not, take the option 1. Work with current main storage dump (MSD), and then hit F11=Copy status.
You can watch the MSD status, including the Storage Management Recovery steps. Be aware that 1-3%, and 99%, cover special functions that can make them take longer than the other percentages. Do not estimate dump copy time from those percentages.
Once the copy completes, exit back to the MSD Manager menu. Do not exit past that, as you could start the IPL back up before we collect the other dump type.
Take option 3. Work with copy options. If the 'Subset if possible' value is set to '1=Yes', as is default at 7.2, change it to '3=No' and hit Enter. If the 'Subset if possible' value is set to '3=No', as is default prior to IBM i 7.2, change it to '1=Yes' and hit Enter.
This will take you back to the MSD Manager menu, and you should see a 'Copy options accepted.' status message.
Take option 1. Work with current main storage dump (MSD), followed by option 3. Copy to ASP
If you changed the 'Subset if possible' value to 3, then put the description 'Forced MSD - Full Copy'.
If you changed the 'Subset if possible' value to 1, then put the description 'Forced MSD - Subset Copy'.
Hit Enter
Since Storage Management Recovery has already completed, then the copy time will only be the time to actually write the dump to disk.
Once the copy completes, you will be put back to the Work with Current Main Storage Dump menu. Hit F12 to exit to the Main Storage Dump Manager menu. Take option 2. Work with copies of main storage dumps to confirm you have two copies of the dump.
*** Do NOT issue a 5 to display either of the dumps, as that will start an indexing process of the compressed dumps, which can take an extended period of time. ***
Once that is confirmed, F12 back to the Main Storage Manager, and then take option 3. Work with copy options. Change the 'Subset if possible' value back to its original value, and hit Enter.
This will take you back to the MSD Manager menu, and you should see a 'Copy options accepted.' status message.
You can now hit F3 twice to start the IPL back up.
Post-IPL, the recommendation is to submit the Subset copy of the dump to IBM first.
Other useful MSD topics
Copy MSD directly to and from IFS file (IBM i 7.2 and later) - http://www-01.ibm.com/support/docview.wss?uid=nas8N1020052
Customer Instructions for Copying a MSD from ASP (MSD Manager) to a Virtual Tape Image (prior to IBM i 7.2) - http://www-01.ibm.com/support/docview.wss?uid=nas8N1018687
Running Advanced Analysis macro SERVICEDOCS against an MSD - MSD Handling for Very Large Systems, See Option 3 to collect servicedocs against the MSD and send it to IBM https://www.ibm.com/support/pages/msd-handling-very-large-systems
Background for using this procedure
Since IBM i v5r2m0, system crashes had the potential to create a subset dump, if the SRC matched known failure types. In those cases, the system could determine which data was needed, or not needed, to diagnose the failure. However, forced MSDs had broadly been left in the "Can't know why you forced a dump." category. So, we preserved a full copy of partition memory. Over the years this approach has become more challenging as partition memory sizes have gone to the multi-Terabyte ranges. Despite compression of the dumps, the collection and transfer of these dumps can require significant time and resources.
At IBM i 7.2, and PTF'd back to IBM i 6.1.1 and 7.1, we have the function to create a subset dump for forced dumps. This can constitute a greater than 90% space savings in some cases, along with reduced collection and transfer times. At IBM i 6.1.1 and 7.1, the function is disabled by default. At IBM i 7.2, it is enabled by default. The concern is that, with the wide variety of reasons for forcing a dump, that subsetting has the potential to omit needed diagnostic data. As stated above, we needed a procedure that allows you to collect both a Full IBM i Main Storage Dump (MSD), and a Subset MSD, maintaining the ability to go back to the Full dump, if the need arises.
Related Information
[{"Type":"MASTER","Line of Business":{"code":"LOB68","label":"Power HW"},"Business Unit":{"code":"BU070","label":"IBM Infrastructure"},"Product":{"code":"SWG60","label":"IBM i"},"ARM Category":[{"code":"a8m0z0000000C4BAAU","label":"IBM i"}],"ARM Case Number":"","Platform":[{"code":"PF012","label":"IBM i"}],"Version":"7.1.0;7.2.0;7.3.0;7.4.0;7.5.0"}]
Was this topic helpful?
Document Information
Modified date:
07 January 2025
UID
nas8N1020058