vitalif
/
vitastor
5
4
Fork 9
Code Issues 1 Pull Requests 2 Actions Releases 69 Wiki Activity

Return EIO instead of EINVAL for incomplete EC objects

hier-failure-domains
Vitaliy Filippov 2023-05-10 01:38:53 +03:00
parent a409598b16
commit 8cac795445
2 changed files with 23 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
src/cluster_client.cpp
View File

@ -743,15 +743,16 @@ resume_3:
erase_op(op); erase_op(op);
return 1; return 1;
} }
else if (op->retval != 0 && op->retval != -EPIPE) else if (op->retval != 0 && op->retval != -EPIPE && op->retval != -EIO && op->retval != -ENOSPC)
{ {
// Fatal error (not -EPIPE) // Fatal error (neither -EPIPE, -EIO nor -ENOSPC)
// FIXME: Add a parameter to allow to not wait for EIOs (incomplete or corrupted objects) to heal
erase_op(op); erase_op(op);
return 1; return 1;
} }
else else
{ {
// -EPIPE - clear the error and retry // Non-fatal error - clear the error and retry
op->retval = 0; op->retval = 0;
if (op->needs_reslice) if (op->needs_reslice)
{ {
@ -1048,7 +1049,7 @@ resume_1:
uw_it->second.state = CACHE_DIRTY; uw_it->second.state = CACHE_DIRTY;
} }
} }
if (op->retval == -EPIPE) if (op->retval == -EPIPE || op->retval == -EIO || op->retval == -ENOSPC)
{ {
// Retry later // Retry later
op->parts.clear(); op->parts.clear();
@ -1119,13 +1120,13 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
{ {
// Operation failed, retry // Operation failed, retry
part->flags |= PART_ERROR; part->flags |= PART_ERROR;
if (!op->retval || op->retval == -EPIPE) if (!op->retval || op->retval == -EPIPE || part->op.reply.hdr.retval == -EIO)
{ {
// Don't overwrite other errors with -EPIPE // Error priority: EIO > ENOSPC > EPIPE
op->retval = part->op.reply.hdr.retval; op->retval = part->op.reply.hdr.retval;
} }
int stop_fd = -1; int stop_fd = -1;
if (op->retval != -EINTR && op->retval != -EIO) if (op->retval != -EINTR && op->retval != -EIO && op->retval != -ENOSPC)
{ {
stop_fd = part->op.peer_fd; stop_fd = part->op.peer_fd;
fprintf( fprintf(
@ -1133,21 +1134,25 @@ void cluster_client_t::handle_op_part(cluster_op_part_t *part)
osd_op_names[part->op.req.hdr.opcode], part->osd_num, part->op.reply.hdr.retval, expected osd_op_names[part->op.req.hdr.opcode], part->osd_num, part->op.reply.hdr.retval, expected
); );
} }
else
{
fprintf(
stderr, "%s operation failed on OSD %lu: retval=%ld (expected %d)\n",
osd_op_names[part->op.req.hdr.opcode], part->osd_num, part->op.reply.hdr.retval, expected
);
}
// All next things like timer, continue_sync/rw and stop_client may affect the operation again // All next things like timer, continue_sync/rw and stop_client may affect the operation again
// So do all these things after modifying operation state, otherwise we may hit reenterability bugs // So do all these things after modifying operation state, otherwise we may hit reenterability bugs
// FIXME postpone such things to set_immediate here to avoid bugs // FIXME postpone such things to set_immediate here to avoid bugs
if (part->op.reply.hdr.retval == -EPIPE) // Mark op->up_wait = true to retry operation after a short pause (not immediately)
op->up_wait = true;
if (!retry_timeout_id)
{ {
// Mark op->up_wait = true before stopping the client retry_timeout_id = tfd->set_timer(up_wait_retry_interval, false, [this](int)
op->up_wait = true;
if (!retry_timeout_id)
{ {
retry_timeout_id = tfd->set_timer(up_wait_retry_interval, false, [this](int) retry_timeout_id = 0;
{ continue_ops(true);
retry_timeout_id = 0; });
continue_ops(true);
});
}
} }
if (op->inflight_count == 0) if (op->inflight_count == 0)
{ {

2
src/osd_primary_write.cpp
View File

@ -81,7 +81,7 @@ resume_1:
if (!cur_op->rmw_buf) if (!cur_op->rmw_buf)
{ {
// Refuse partial overwrite of an incomplete object // Refuse partial overwrite of an incomplete object
cur_op->reply.hdr.retval = -EINVAL; cur_op->reply.hdr.retval = -EIO;
goto continue_others; goto continue_others;
} }
} }