diff options
author | Pranith Kumar K <pkarampu@redhat.com> | 2017-01-25 15:31:44 +0530 |
---|---|---|
committer | Pranith Kumar Karampuri <pkarampu@redhat.com> | 2017-02-26 22:06:55 -0500 |
commit | c1fc1fc9cb5a13e6ddf8c9270deb0c7609333540 (patch) | |
tree | a3876aa8a0c1b087429ba916c9380b90bcda6b72 /xlators/cluster/ec/src/ec-common.c | |
parent | 4638dfc1fee80f9338f2941f3cccb17bec63989a (diff) |
cluster/ec: Don't trigger data/metadata heal on Lookups
Problem-1
If Lookup which doesn't take any locks observes version mismatch it can't be
trusted. If we launch a heal based on this information it will lead to
self-heals which will affect I/O performance in the cases where Lookup is
wrong. Considering self-heal-daemon and operations on the inode from client
which take locks can still trigger heal we can choose to not attempt a heal on
Lookup.
Problem-2:
Fixed spurious failure of
tests/bitrot/bug-1373520.t
For the issues above, what was happening was that ec_heal_inspect()
is preventing 'name' heal to happen
Problem-3:
tests/basic/ec/ec-background-heals.t
To be honest I don't know what the problem was, while fixing
the 2 problems above, I made some changes to ec_heal_inspect() and
ec_need_heal() after which when I tried to recreate the spurious
failure it just didn't happen even after a long time.
BUG: 1414287
Signed-off-by: Pranith Kumar K <pkarampu@redhat.com>
Change-Id: Ife2535e1d0b267712973673f6d474e288f3c6834
Reviewed-on: https://review.gluster.org/16468
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Xavier Hernandez <xhernandez@datalab.es>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Ashish Pandey <aspandey@redhat.com>
Diffstat (limited to 'xlators/cluster/ec/src/ec-common.c')
-rw-r--r-- | xlators/cluster/ec/src/ec-common.c | 66 |
1 files changed, 52 insertions, 14 deletions
diff --git a/xlators/cluster/ec/src/ec-common.c b/xlators/cluster/ec/src/ec-common.c index 5e1efe381d4..823922542a0 100644 --- a/xlators/cluster/ec/src/ec-common.c +++ b/xlators/cluster/ec/src/ec-common.c @@ -82,10 +82,50 @@ int32_t ec_heal_report(call_frame_t * frame, void * cookie, xlator_t * this, return 0; } +static uintptr_t +ec_fop_needs_name_heal (ec_fop_data_t *fop) +{ + ec_t *ec = NULL; + ec_cbk_data_t *cbk = NULL; + ec_cbk_data_t *enoent_cbk = NULL; + + ec = fop->xl->private; + if (fop->id != GF_FOP_LOOKUP) + return 0; + + if (!fop->loc[0].name || strlen (fop->loc[0].name) == 0) + return 0; + + list_for_each_entry(cbk, &fop->cbk_list, list) + { + if (cbk->op_ret < 0 && cbk->op_errno == ENOENT) { + enoent_cbk = cbk; + break; + } + } + + if (!enoent_cbk) + return 0; + + return ec->xl_up & ~enoent_cbk->mask; +} + int32_t ec_fop_needs_heal(ec_fop_data_t *fop) { ec_t *ec = fop->xl->private; + if (fop->lock_count == 0) { + /* + * if fop->lock_count is zero that means it saw version mismatch + * without any locks so it can't be trusted. If we launch a heal + * based on this it will lead to INODELKs which will affect I/O + * performance. Considering self-heal-daemon and operations on + * the inode from client which take locks can still trigger the + * heal we can choose to not attempt a heal when fop->lock_count + * is zero. + */ + return 0; + } return (ec->xl_up & ~(fop->remaining | fop->good)) != 0; } @@ -95,7 +135,7 @@ void ec_check_status(ec_fop_data_t * fop) int32_t partial = 0; char str1[32], str2[32], str3[32], str4[32], str5[32]; - if (!ec_fop_needs_heal(fop)) { + if (!ec_fop_needs_name_heal (fop) && !ec_fop_needs_heal(fop)) { return; } @@ -108,19 +148,17 @@ void ec_check_status(ec_fop_data_t * fop) } } - if (fop->lock_count > 0) { - gf_msg (fop->xl->name, GF_LOG_WARNING, 0, - EC_MSG_OP_FAIL_ON_SUBVOLS, - "Operation failed on %d of %d subvolumes.(up=%s, mask=%s, " - "remaining=%s, good=%s, bad=%s)", - gf_bits_count(ec->xl_up & ~(fop->remaining | fop->good)), ec->nodes, - ec_bin(str1, sizeof(str1), ec->xl_up, ec->nodes), - ec_bin(str2, sizeof(str2), fop->mask, ec->nodes), - ec_bin(str3, sizeof(str3), fop->remaining, ec->nodes), - ec_bin(str4, sizeof(str4), fop->good, ec->nodes), - ec_bin(str5, sizeof(str5), - ec->xl_up & ~(fop->remaining | fop->good), ec->nodes)); - } + gf_msg (fop->xl->name, GF_LOG_WARNING, 0, + EC_MSG_OP_FAIL_ON_SUBVOLS, + "Operation failed on %d of %d subvolumes.(up=%s, mask=%s, " + "remaining=%s, good=%s, bad=%s)", + gf_bits_count(ec->xl_up & ~(fop->remaining | fop->good)), ec->nodes, + ec_bin(str1, sizeof(str1), ec->xl_up, ec->nodes), + ec_bin(str2, sizeof(str2), fop->mask, ec->nodes), + ec_bin(str3, sizeof(str3), fop->remaining, ec->nodes), + ec_bin(str4, sizeof(str4), fop->good, ec->nodes), + ec_bin(str5, sizeof(str5), + ec->xl_up & ~(fop->remaining | fop->good), ec->nodes)); if (fop->use_fd) { if (fop->fd != NULL) { |