The VM can fail normal allocations quite often under severe load. The
VFS just ignores this in getblk, retrying silently, but ext3 emits a
warning; the patch just makes them a bit less frequent.
--- linux-2.4-ext3push/fs/jbd/journal.c.=K0003=.orig 2003-03-25 10:59:15.000000000 +0000
+++ linux-2.4-ext3push/fs/jbd/journal.c 2003-03-25 10:59:15.000000000 +0000
@@ -1660,7 +1660,7 @@ void * __jbd_kmalloc (char *where, size_
* messages. */
jbd_debug(1, "ENOMEM in %s, retrying.\n", where);
- if (time_after(jiffies, last_warning + 5*HZ)) {
+ if (time_after(jiffies, last_warning + 120*HZ)) {
printk(KERN_NOTICE
"ENOMEM in %s, retrying.\n", where);
last_warning = jiffies;