 |
 |
Here is something else I have discovered.
The buffer/chunk size used by the xrdcp client is 8388608, not 4194304. When it talks to the dCache server, it gets 3 oksofar + 1 ok for each read request, with the final remainder being just ok:
[2021-08-06 08:23:35.360875 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_oksofar response to request kXR_read (handle: 0x00000000, offset: 981467136, size: 8388608)
[2021-08-06 08:23:35.378615 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_oksofar response to request kXR_read (handle: 0x00000000, offset: 981467136, size: 8388608)
[2021-08-06 08:23:35.396493 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_oksofar response to request kXR_read (handle: 0x00000000, offset: 981467136, size: 8388608)
[2021-08-06 08:23:35.432073 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_oksofar response to request kXR_read (handle: 0x00000000, offset: 989855744, size: 8388608)
[2021-08-06 08:23:35.432074 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 981467136, size: 8388608)
[2021-08-06 08:23:35.449904 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_oksofar response to request kXR_read (handle: 0x00000000, offset: 989855744, size: 8388608)
[2021-08-06 08:23:35.467742 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_oksofar response to request kXR_read (handle: 0x00000000, offset: 989855744, size: 8388608)
[2021-08-06 08:23:35.503334 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 989855744, size: 8388608)
[2021-08-06 08:23:35.518253 -0500][Dump ][XRootD ] [dcatest04.fnal.gov:33138] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 998244352, size: 1755648)
(the out-of-order handling is of not immediate consequence)
now, with RAL, the 8388608 buffer provokes NO kXR_oksofar responses! (but the 4194304 buffer we use did!):
[2021-08-06 08:04:29.480545 -0500][Dump ][XRootD ] [ceph-gw8.gridpp.rl.ac.uk:1094] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 964689920, size: 8388608)
[2021-08-06 08:04:30.328868 -0500][Dump ][XRootD ] [ceph-gw8.gridpp.rl.ac.uk:1094] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 973078528, size: 8388608)
[2021-08-06 08:04:31.462947 -0500][Dump ][XRootD ] [ceph-gw8.gridpp.rl.ac.uk:1094] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 981467136, size: 8388608)
[2021-08-06 08:04:32.499131 -0500][Dump ][XRootD ] [ceph-gw8.gridpp.rl.ac.uk:1094] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 989855744, size: 8388608)
[2021-08-06 08:04:32.692708 -0500][Dump ][XRootD ] [ceph-gw8.gridpp.rl.ac.uk:1094] Got a kXR_ok response to request kXR_read (handle: 0x00000000, offset: 998244352, size: 1755648)
So clearly the reason xrdcp is not failing is simply because it is not getting the strange kXR_oksofar that we are.
I am curious if increasing the chunk size actually solves this. I will try that out.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or unsubscribe.
Triage notifications on the go with GitHub Mobile for iOS or Android.
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-DEV list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-DEV&A=1