Hi, during some stress tests with lustre-1.4.9 we got error messages and don''t understand their meanings. Client: ======[15016.022175] LustreError: 3076:0:(client.c:576:ptlrpc_check_status()) @@@ type == PTL_RPC_MSG_ERR, err == -28 req@ffff8101327668f8 x140234/t0 o4->ost2_UUID@beo-102_UUID:28 lens 328/288 ref 2 fl Rpc:R/0/0 rc 0/-28 [44739.193858] LustreError: 3076:0:(client.c:576:ptlrpc_check_status()) @@@ type == PTL_RPC_MSG_ERR, err == -28 req@ffff8101318cc118 x421028/t0 o4->ost2_UUID@beo-102_UUID:28 lens 328/288 ref 2 fl Rpc:R/0/0 rc 0/-28 MDS: ===[303033.100430] LustreError: 4253:0: (lov_request.c:666:lov_update_create_set()) error creating fid 0xbe2e5e sub-object on OST idx 1/1: rc = -28 [303037.518295] LustreError: 4260:0:(mds_open.c:714:mds_finish_open()) mds_create_objects: rc = -28 During the tests, the iozone test jobs required more disk space than available, maybe the messages above correlate to this problem? Unfortunately klogd was not running, so I''m not sure about it. Thanks in advance, Bernd -- Bernd Schubert Q-Leap Networks GmbH / transtec AG / ttec
Hi, On Tue, 2007-03-20 at 11:42 +0100, Bernd Schubert wrote:> Hi, > > during some stress tests with lustre-1.4.9 we got error messages and don''t > understand their meanings. > > Client: > ======> [15016.022175] LustreError: 3076:0:(client.c:576:ptlrpc_check_status()) @@@ > type == PTL_RPC_MSG_ERR, err == -28 req@ffff8101327668f8 x140234/t0 > o4->ost2_UUID@beo-102_UUID:28 lens 328/288 ref 2 fl Rpc:R/0/0 rc 0/-28 > > [44739.193858] LustreError: 3076:0:(client.c:576:ptlrpc_check_status()) @@@ > type == PTL_RPC_MSG_ERR, err == -28 req@ffff8101318cc118 x421028/t0 > o4->ost2_UUID@beo-102_UUID:28 lens 328/288 ref 2 fl Rpc:R/0/0 rc 0/-28 >-28 = -ENOSPC (per /usr/include/asm/errno.h) Seems as if OST2 is full. Regards, Kalpak.> > MDS: > ===> [303033.100430] LustreError: 4253:0: > (lov_request.c:666:lov_update_create_set()) error creating fid 0xbe2e5e > sub-object on OST idx 1/1: rc = -28 > > [303037.518295] LustreError: 4260:0:(mds_open.c:714:mds_finish_open()) > mds_create_objects: rc = -28 > > > During the tests, the iozone test jobs required more disk space than > available, maybe the messages above correlate to this problem? Unfortunately > klogd was not running, so I''m not sure about it. > > > Thanks in advance, > Bernd >
On Tuesday 20 March 2007 12:28, Kalpak Shah wrote:> > [44739.193858] LustreError: 3076:0:(client.c:576:ptlrpc_check_status()) > > @@@ type == PTL_RPC_MSG_ERR, err == -28 req@ffff8101318cc118 x421028/t0 > > o4->ost2_UUID@beo-102_UUID:28 lens 328/288 ref 2 fl Rpc:R/0/0 rc 0/-28 > > -28 = -ENOSPC (per /usr/include/asm/errno.h)>Seems as if OST2 is full.Ah, thanks a lot, I will look into this file next time. -- Bernd Schubert Q-Leap Networks GmbH / transtec AG / ttec