In the current implementation, parallel dump won't perform well in --num-threads 1 or -d 31. So this patch add the description of it.
Signed-off-by: Zhou Wenjian zhouwj-fnst@cn.fujitsu.com --- kexec-kdump-howto.txt | 3 +++ 1 file changed, 3 insertions(+)
diff --git a/kexec-kdump-howto.txt b/kexec-kdump-howto.txt index b4cdc22..00ab96b 100644 --- a/kexec-kdump-howto.txt +++ b/kexec-kdump-howto.txt @@ -627,6 +627,9 @@ For example:
1 has better performance than 2, if THREAD_NUM is larger than two and the usable cpus number is larger than THREAD_NUM. +Limitation: Parallel feature will introduce extra costs to each page(including +filtered page). So it will always have worse performance when the THREAD_NUM +is 1 or -d 31 is specified.
Notes on how to use multiple cpus on a capture kernel on x86 system: