Description
Increasing the size of a ZFS volume used as a swap area or making a new ZFS volume available as a swap area may cause Solaris 10 systems (with certain patches installed) and Solaris 11 systems to panic if there is an attempt to use the newly available swap area within a minute of the new swap area being configured.
Occurrence
This issue can occur in the following releases:
SPARC Platform
•Solaris 10 with patch 150400-06 or later
•Solaris 11.1.13.6.0 or later
x86 Platform
•Solaris 10 with patch 150401-06 or later
•Solaris 11.1.13.6.0 or later
Note 1: Solaris 8 and Solaris 9 are not impacted by this issue.
Note 2: At the time of the initial publication of this document, the Solaris 10 patches 150400-06 and 150401-06 were not yet released to MOS.
Symptoms
If the described issue occurs, the system will panic with a stack trace similar to the following:
zvol_preallocio+0x19b(ffffc10007cbfac0, 0, de7c000, 1000, 40, ffffc10004b77580)
zvol_strategy+0x2ec(ffffc10004b77580)
bdev_strategy+0x66(ffffc10004b77580)
spec_startio+0xa4(ffffc10004d55800, ffffe33000bd8000, de7c000, 1000, 40)
spec_pageio+0x3c(ffffc10004d55800, ffffe33000bd8000, de7c000, 1000, 40, ffffc100132e9280)
fop_pageio+0xba(ffffc10004d55800, ffffe33000bd8000, de7c000, 1000, 40, ffffc100132e9280)
swap_getapage+0x217(ffffc10004862880, 1ffff8200128f000, 1000, fffffffc8034db30, fffffffc8034dab0, 1000)
swap_getpage+0x90(ffffc10004862880, 1ffff8200128f000, 1000, fffffffc8034db30, fffffffc8034dab0, 1000)
fop_getpage+0xc8(ffffc10004862880, 1ffff8200128f000, 1000, fffffffc8034db30, fffffffc8034dab0, 1000)
anon_getpage+0x174(fffffffc8034da80, fffffffc8034db30, fffffffc8034dab0, 1000, ffffc1000ffafb98, 8a7d19000)
segvn_faultpage+0x1b6(ffffc10000071218, ffffc1000ffafb98, 8a7d19000, f1c000, 0, fffffffc8034dc00)
segvn_fault+0xb78(ffffc10000071218, ffffc1000ffafb98, 8a7d19000, 1000, 0, 2)
as_fault+0x3e7(ffffc10000071218, ffffc100084d4d30, 8a7d19320, 1, 0, 2)
pagefault+0x99(8a7d19320, 0, 2, 0)
trap+0xce0(fffffffc8034df10, 8a7d19320, 0)
0xfffffffffb8001d7()
Workaround
To avoid this issue, do not add a new swap ZFS volume to the swap device or increase the size of an existing one if the system is running out of physical memory. Instead, stop memory intensive applications, increase the size of existing swap area or add a new one, wait one minute and then resume the normal operations.
Increasing the size of a ZFS volume used as a swap area or making a new ZFS volume available as a swap area may cause Solaris 10 systems (with certain patches installed) and Solaris 11 systems to panic if there is an attempt to use the newly available swap area within a minute of the new swap area being configured.
Occurrence
This issue can occur in the following releases:
SPARC Platform
•Solaris 10 with patch 150400-06 or later
•Solaris 11.1.13.6.0 or later
x86 Platform
•Solaris 10 with patch 150401-06 or later
•Solaris 11.1.13.6.0 or later
Note 1: Solaris 8 and Solaris 9 are not impacted by this issue.
Note 2: At the time of the initial publication of this document, the Solaris 10 patches 150400-06 and 150401-06 were not yet released to MOS.
Symptoms
If the described issue occurs, the system will panic with a stack trace similar to the following:
zvol_preallocio+0x19b(ffffc10007cbfac0, 0, de7c000, 1000, 40, ffffc10004b77580)
zvol_strategy+0x2ec(ffffc10004b77580)
bdev_strategy+0x66(ffffc10004b77580)
spec_startio+0xa4(ffffc10004d55800, ffffe33000bd8000, de7c000, 1000, 40)
spec_pageio+0x3c(ffffc10004d55800, ffffe33000bd8000, de7c000, 1000, 40, ffffc100132e9280)
fop_pageio+0xba(ffffc10004d55800, ffffe33000bd8000, de7c000, 1000, 40, ffffc100132e9280)
swap_getapage+0x217(ffffc10004862880, 1ffff8200128f000, 1000, fffffffc8034db30, fffffffc8034dab0, 1000)
swap_getpage+0x90(ffffc10004862880, 1ffff8200128f000, 1000, fffffffc8034db30, fffffffc8034dab0, 1000)
fop_getpage+0xc8(ffffc10004862880, 1ffff8200128f000, 1000, fffffffc8034db30, fffffffc8034dab0, 1000)
anon_getpage+0x174(fffffffc8034da80, fffffffc8034db30, fffffffc8034dab0, 1000, ffffc1000ffafb98, 8a7d19000)
segvn_faultpage+0x1b6(ffffc10000071218, ffffc1000ffafb98, 8a7d19000, f1c000, 0, fffffffc8034dc00)
segvn_fault+0xb78(ffffc10000071218, ffffc1000ffafb98, 8a7d19000, 1000, 0, 2)
as_fault+0x3e7(ffffc10000071218, ffffc100084d4d30, 8a7d19320, 1, 0, 2)
pagefault+0x99(8a7d19320, 0, 2, 0)
trap+0xce0(fffffffc8034df10, 8a7d19320, 0)
0xfffffffffb8001d7()
Workaround
To avoid this issue, do not add a new swap ZFS volume to the swap device or increase the size of an existing one if the system is running out of physical memory. Instead, stop memory intensive applications, increase the size of existing swap area or add a new one, wait one minute and then resume the normal operations.
No comments:
Post a Comment