|  |  | 
  
    |  |  | 
 
  |   |  | 
  
    |  |  | 
  
    |  |  | 
  
    |   xen-ppc-devel
Re: [XenPPC] making xen tools on dom0 can be slow 
| 
Maria Butrico wrote:
 Jimi and I have looked more carefully at the xend startup problem.  xend 
is very slow if the tools or any part of the environment variable PATH 
are in an nfs mounted directory.  Xend starts in reasonable time if the 
tools are local and no part of the PATH environment variable points a to 
a remote directory.
This was discussed yesterday and it was suggested that I should enter a bug
for this.  Since then I have done a few experiments:
By default dom0 has 64M.  Under these condition make tools takes 24 minutes
on a js20.
make[1]: Leaving directory `/home/j9/xen-ppc/tools'
real    24m1.321s
user    0m34.716s
sys     0m23.272s
I have found this is about the case on average.  This is with the xen
source tree on nfs.  I dismiss this as a network related problem since I
see no evidence of this.
By changing xen to give 128M to each domain and therefore to dom0, the same
make, also with source over nfs takes less than 2 minutes.
make[1]: Leaving directory `/home/j9/xen-ppc/tools'
real    1m15.445s
user    0m32.800s
sys     0m14.753s
I will not enter this as a bug.  If anything the bug is that one might not
specify the memory size for dom0.
(On a related note, once we alter xen to give each domain 128M, xm create
no longer works.  Jimi is debugging this.)
Also I reported slowness for starting xend.  I am still working with this
and will post to the list as soon as I have more data.  Simply giving more
memory to dom0 does not fix the slow xend start problem.
Maria Butrico    <internet or sametime: butrico@xxxxxxxxxx;     Notes:
Maria Butrico/Watson/IBM>
_______________________________________________
Xen-ppc-devel mailing list
Xen-ppc-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ppc-devel
 
_______________________________________________
Xen-ppc-devel mailing list
Xen-ppc-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ppc-devel
 | 
 |  | 
  
    |  |  |