HPSS Tickets: Difference between revisions

From Lsdf
Jump to navigationJump to search
No edit summary
No edit summary
Line 11: Line 11:
!Description and History
!Description and History
|-
|-
|HPSS
|Hardware
|open
|open
|CR2803
|01VV2YN,724
|2015-07-15
|2015-10-14
by Nathan
by Scott
|<nowiki>-</nowiki>
|<nowiki>-</nowiki>
|'''Zero byte files in HPSS should have data'''
|'''DCS3700+ T10PI partitioning'''
While changing the COS, it was seen that zero byte files were not migrated to the new COS. Taking a closer look, some of these files (~20k) should not have had zero size. Investigating.
In preparation for E2EDI (end to end data integrity), the arrays need to be reformatted with T10PI option enabled. This was initially planned for the system setup, but back then the DCS firmware had a bug which made formatting with T10PI extremely slow.
Ticket was opened to ask in which firmware version this bug was fixed.
<br />''2015-07-16:'' No bug known related to slow T10PI formatting - it just takes that long.
<br />''2015-07-29:'' Reopened the ticket, question about parallelism in formatting LUNs.
|-
|Hardware
|open
|01VV2MW,724
|2015-07-16
by Nathan
|<nowiki>-</nowiki>
|'''Mover Server RAID Adapter'''
In preparation to the bare metal restore testing, we need to know how the M5110 RAID adapter behaves in different configurations.
|}
|}


Line 63: Line 51:
Discussing change of SC/COS layout. Assuming there's a need to change the storage class for files that have already been written to tape. The new SC would use the same kind of tape media with the same settings. Is it still necessary to move data physically to new tapes when changing SC for these files?
Discussing change of SC/COS layout. Assuming there's a need to change the storage class for files that have already been written to tape. The new SC would use the same kind of tape media with the same settings. Is it still necessary to move data physically to new tapes when changing SC for these files?
<br />''2015-07-02:'' Unfortunately, yes. There is a command to change a file's SC, but it inevitably includes moving the data.
<br />''2015-07-02:'' Unfortunately, yes. There is a command to change a file's SC, but it inevitably includes moving the data.
|-
|Hardware
|closed
|01VV2YN,724
|2015-07-15
by Nathan
|2015-07-27
|'''DCS3700+ T10PI partitioning'''
In preparation for E2EDI (end to end data integrity), the arrays need to be reformatted with T10PI option enabled. This was initially planned for the system setup, but back then the DCS firmware had a bug which made formatting with T10PI extremely slow.
Ticket was opened to ask in which firmware version this bug was fixed.
<br />''2015-07-16:'' No bug known related to slow T10PI formatting - it just takes that long.
<br />''2015-07-29:'' Reopened the ticket, question about parallelism in formatting LUNs. Answer: there are default settings in the firmware, optimized for hardware. Difference between DCS3700 and DCS3700+. May be possible to change values, but not recommended. Ticket closed again.
|-
|Hardware
|closed
|01VV2MW,724
|2015-07-16
by Nathan
|2015-07-16
|'''Mover Server RAID Adapter'''
In preparation to the bare metal restore testing, we need to know how the M5110 RAID adapter behaves in different configurations.
<br />''2015-07-16:'' RAID config is stored on all disks that are part of the RAID. Using the RAID config tool, the user can decide to restore settings from disk (if available), which will keep data on disks that were part of that RAID array before.
|}
|}

Revision as of 12:03, 16 October 2015

Tracking list for current and past tickets related to HPSS.

Open Tickets

Component Status Ticket ID Opened Closed Description and History
HPSS open CR2803 2015-10-14

by Scott

- Zero byte files in HPSS should have data

While changing the COS, it was seen that zero byte files were not migrated to the new COS. Taking a closer look, some of these files (~20k) should not have had zero size. Investigating.

Past Tickets

Please keep sorted by opening date.

Component Status Ticket ID Opened Closed Description and History
HPSS closed CR2710 2015-06-23

by Nathan

2015-07-02 Core Server System Backup and DB2

Which of partitions containing DB2 logs need to be included in a backup? What about accounting logs?
2015-07-02: Backing up log archives is sufficient. Accounting data is included in the DB2 backup. Accounting logs can be recreated from there, if necessary.

HPSS closed CR2711 2015-06-23

by Nathan

2015-07-02 Changing SC for Files on Tape

Discussing change of SC/COS layout. Assuming there's a need to change the storage class for files that have already been written to tape. The new SC would use the same kind of tape media with the same settings. Is it still necessary to move data physically to new tapes when changing SC for these files?
2015-07-02: Unfortunately, yes. There is a command to change a file's SC, but it inevitably includes moving the data.

Hardware closed 01VV2YN,724 2015-07-15

by Nathan

2015-07-27 DCS3700+ T10PI partitioning

In preparation for E2EDI (end to end data integrity), the arrays need to be reformatted with T10PI option enabled. This was initially planned for the system setup, but back then the DCS firmware had a bug which made formatting with T10PI extremely slow. Ticket was opened to ask in which firmware version this bug was fixed.
2015-07-16: No bug known related to slow T10PI formatting - it just takes that long.
2015-07-29: Reopened the ticket, question about parallelism in formatting LUNs. Answer: there are default settings in the firmware, optimized for hardware. Difference between DCS3700 and DCS3700+. May be possible to change values, but not recommended. Ticket closed again.

Hardware closed 01VV2MW,724 2015-07-16

by Nathan

2015-07-16 Mover Server RAID Adapter

In preparation to the bare metal restore testing, we need to know how the M5110 RAID adapter behaves in different configurations.
2015-07-16: RAID config is stored on all disks that are part of the RAID. Using the RAID config tool, the user can decide to restore settings from disk (if available), which will keep data on disks that were part of that RAID array before.