From: liuy <liuy>
Date: Thu, 16 Aug 2007 14:12:20 +0000 (+0000)
Subject: Branch b1_6
X-Git-Tag: v1_8_0_110~1328
X-Git-Url: https://git.whamcloud.com/gitweb?a=commitdiff_plain;h=9d99b9572ba085e2d85bf877c11ce00fdeb93540;p=fs%2Flustre-release.git

Branch b1_6
b=12932
i=green, shadow

set obd_health_check_timeout as 1.5x of obd_timeout
---

diff --git a/lustre/ChangeLog b/lustre/ChangeLog
index f0b8820..9d03c7f 100644
--- a/lustre/ChangeLog
+++ b/lustre/ChangeLog
@@ -14,6 +14,11 @@
        * Recommended e2fsprogs version: 1.40.2-cfs1
        * Note that reiserfs quotas are disabled on SLES 10 in this kernel.
 
+Severity   : major
+Bugzilla   : 12932
+Description: obd_health_check_timeout too short
+Details    : set obd_health_check_timeout as 1.5x of obd_timeout
+
 2007-08-27         Cluster File Systems, Inc. <info@clusterfs.com>
        * version 1.6.2
        * Support for kernels:
diff --git a/lustre/include/obd_support.h b/lustre/include/obd_support.h
index 875417f..e91a941 100644
--- a/lustre/include/obd_support.h
+++ b/lustre/include/obd_support.h
@@ -37,7 +37,6 @@ extern unsigned int obd_dump_on_eviction;
 extern unsigned int obd_timeout;          /* seconds */
 #define PING_INTERVAL max(obd_timeout / 4, 1U)
 #define RECONNECT_INTERVAL max(obd_timeout / 10, 10U)
-#define LDLM_TIMEOUT_DEFAULT 20
 extern unsigned int ldlm_timeout;
 extern unsigned int obd_health_check_timeout;
 extern unsigned int obd_sync_filter;
@@ -47,6 +46,12 @@ extern cfs_waitq_t obd_race_waitq;
 extern int obd_race_state;
 extern unsigned int obd_alloc_fail_rate;
 
+/* Timeout definitions */
+#define LDLM_TIMEOUT_DEFAULT 20
+#define OBD_TIMEOUT_DEFAULT 100
+#define HEALTH_CHECK_COEF 3 / 2
+#define HEALTH_CHECK_TIMEOUT_DEFAULT (OBD_TIMEOUT_DEFAULT * HEALTH_CHECK_COEF)
+#define HEALTH_CHECK_TIMEOUT (obd_timeout * HEALTH_CHECK_COEF)
 
 #define OBD_FAIL_MDS                     0x100
 #define OBD_FAIL_MDS_HANDLE_UNPACK       0x101
diff --git a/lustre/obdclass/class_obd.c b/lustre/obdclass/class_obd.c
index 76a65e2..fbfcd09 100644
--- a/lustre/obdclass/class_obd.c
+++ b/lustre/obdclass/class_obd.c
@@ -63,9 +63,9 @@ unsigned int obd_alloc_fail_rate;
 unsigned int obd_debug_peer_on_timeout;
 unsigned int obd_dump_on_timeout;
 unsigned int obd_dump_on_eviction;
-unsigned int obd_timeout = 100; /* seconds */
+unsigned int obd_timeout = OBD_TIMEOUT_DEFAULT; /* seconds */
 unsigned int ldlm_timeout = LDLM_TIMEOUT_DEFAULT; /* seconds */
-unsigned int obd_health_check_timeout = 120; /* seconds */
+unsigned int obd_health_check_timeout = HEALTH_CHECK_TIMEOUT_DEFAULT; /* seconds */
 unsigned int obd_max_dirty_pages = 256;
 atomic_t obd_dirty_pages;
 
diff --git a/lustre/obdclass/obd_config.c b/lustre/obdclass/obd_config.c
index a1d9fd7..4c01e23 100644
--- a/lustre/obdclass/obd_config.c
+++ b/lustre/obdclass/obd_config.c
@@ -760,6 +760,7 @@ int class_process_config(struct lustre_cfg *lcfg)
                 CDEBUG(D_IOCTL, "changing lustre timeout from %d to %d\n",
                        obd_timeout, lcfg->lcfg_num);
                 obd_timeout = max(lcfg->lcfg_num, 1U);
+                obd_health_check_timeout = HEALTH_CHECK_TIMEOUT;
                 GOTO(out, err = 0);
         }
         case LCFG_SET_UPCALL: {