Hi Thomas,
I don't have a real answer for that. Obviously, some data had to be read
into the cache. I seems to be the way "read into cache" is calculate and
is sensitive to the actual read size. So, I guess you could classify it as
a bug. The other statistics seem OK.
Andy
On Wed, 4 Mar 2020, Thomas Byrne - UKRI STFC wrote:
> Hi,
>
> Thanks for this Andy, it's very helpful.
>
> I'm confused why 'Read' is *always* zero on our cache log stats, I would have expected it to be a multiple of the pagesize. Have I got the wrong end of the stick here?
>
> Cheers
> Tom
>
>> -----Original Message-----
>> From: [log in to unmask] <[log in to unmask]> On
>> Behalf Of Andrew Hanushevsky
>> Sent: 02 March 2020 23:59
>> To: Patargias, George (STFC,RAL,SC) <[log in to unmask]>
>> Cc: [log in to unmask]
>> Subject: Re: Meaning of memory cacheing log stats
>>
>> Hi George,
>>
>> So, here it is:
>>
>> 0 Read; Total number of bytes read into the cache
>> 38581 Get; Number of bytes delivered from the cache
>> 0 Pass; Number of bytes read but not cached
>> 0 Write; Total number of bytes written from the cache
>> 0 Put; Number of bytes updated in the cache
>> 20 Hits; Number of times wanted data was in the cache
>> 3 Miss; Number of times wanted data was *not* in the cache
>> 0 pead; Bytes read via preread (not included in BytesRead)
>> 0 HitsPR; Number of pages of wanted data was just preread
>> 0 MissPR Number of pages of unwanted data was just preread
>>
>> So, it would seem the cache is being effective as number of hits is far greater
>> than the number of misses.
>>
>> Andy
>>
>>
>> On Mon, 2 Mar 2020, George Patargias - UKRI STFC wrote:
>>
>>> Hello,
>>>
>>>
>>> We have turned on the printing of mem cache statistics using the
>>> "logstasts" flag of the pss.cache
>>>
>>> directive as shown below
>>>
>>>
>>> pss.cache logstats max2cache 33554432 pagesize 33554432 size
>>> 17179869184
>>>
>>>
>>> Can you please explain briefly what is the meaning of the numbers
>> appearing in the corresponding lines of the xrootd log? The following is a
>> common example we see on our xrootd server but it is not clear whether the
>> cache is being effective in this case (we suspect it isn't).
>>>
>>>
>>> Cache: Stats: 0 Read; 38581 Get; 0 Pass; 0 Write; 0 Put; 20 Hits; 3
>> Miss; 0 pead; 0 HitsPR; 0 MissPR; Path
>>>
>>>
>>> Thanks,
>>>
>>>
>>> George
>>>
>>>
>> ############################################################
>> ##########
>>> ##
>>> Use REPLY-ALL to reply to list
>>>
>>> To unsubscribe from the XROOTD-L list, click the following link:
>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>>>
>>
>> ############################################################
>> ############
>> Use REPLY-ALL to reply to list
>>
>> To unsubscribe from the XROOTD-L list, click the following link:
>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
|