|
In this Document
APPLIES TO:Linux OS - Version Oracle Linux 6.0 and laterLinux x86-64 SYMPTOMSRunning Oracle RDBM 11.2.0.4 PSU4and CRS on Oracle Linux 6.6 and when instance starts and mounted orarootagent.bin uses a large amount of cpu and also high load. The issue could be reproduced with UEK2/UEK3/RHCK when Oracle instance is started and DB is mounted. Due to this issue,lot of cpu and load on the system is extremely high,cannot run the database. CAUSEThe server has large number of LUNs/paths,and ASM volumes are managered by UDEV rules not ASMLib. udevadm monitor shows when cpu usage spiking,there are lots udev changes.
UDEV [1427349083.901194] change /devices/virtual/block/dm-107 (block)
UDEV [1427349083.902318] change /devices/virtual/block/dm-47 (block) UDEV [1427349083.902425] change /devices/virtual/block/dm-165 (block) UDEV [1427349083.903671] change /devices/virtual/block/dm-56 (block) UDEV [1427349083.904231] change /devices/virtual/block/dm-116 (block) UDEV [1427349083.905124] change /devices/virtual/block/dm-66 (block) UDEV [1427349083.912189] change /devices/virtual/block/dm-93 (block) UDEV [1427349083.913394] change /devices/virtual/block/dm-105 (block) UDEV [1427349083.915191] change /devices/virtual/block/dm-123 (block) UDEV [1427349083.916158] change /devices/virtual/block/dm-170 (block) UDEV [1427349083.920331] change /devices/virtual/block/dm-140 (block) UDEV [1427349083.922613] change /devices/virtual/block/dm-244 (block) UDEV [1427349083.928483] change /devices/virtual/block/dm-213 (block) SOLUTION1. Disable cpuspeed service 2. Limit udev child creation by:
# udevadm --children-max=32
REFERENCES |
|
In this Document
APPLIES TO:Linux OS - Version Oracle Linux 6.0 and laterLinux x86-64 SYMPTOMSRunning Oracle RDBM 11.2.0.4 PSU4and CRS on Oracle Linux 6.6 and when instance starts and mounted orarootagent.bin uses a large amount of cpu and also high load. The issue could be reproduced with UEK2/UEK3/RHCK when Oracle instance is started and DB is mounted. Due to this issue,cannot run the database. CAUSEThe server has large number of LUNs/paths,and ASM volumes are managered by UDEV rules not ASMLib. udevadm monitor shows when cpu usage spiking,there are lots udev changes.
UDEV [1427349083.901194] change /devices/virtual/block/dm-107 (block)
UDEV [1427349083.902318] change /devices/virtual/block/dm-47 (block) UDEV [1427349083.902425] change /devices/virtual/block/dm-165 (block) UDEV [1427349083.903671] change /devices/virtual/block/dm-56 (block) UDEV [1427349083.904231] change /devices/virtual/block/dm-116 (block) UDEV [1427349083.905124] change /devices/virtual/block/dm-66 (block) UDEV [1427349083.912189] change /devices/virtual/block/dm-93 (block) UDEV [1427349083.913394] change /devices/virtual/block/dm-105 (block) UDEV [1427349083.915191] change /devices/virtual/block/dm-123 (block) UDEV [1427349083.916158] change /devices/virtual/block/dm-170 (block) UDEV [1427349083.920331] change /devices/virtual/block/dm-140 (block) UDEV [1427349083.922613] change /devices/virtual/block/dm-244 (block) UDEV [1427349083.928483] change /devices/virtual/block/dm-213 (block) SOLUTION1. Disable cpuspeed service 2. Limit udev child creation by:
# udevadm --children-max=32
REFERENCES |
