UPDATE: I was a little negative here. As I filed the bug report and went over all the details, I think I found the solution. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=250770
-----------------
Hi, I've been with FreeBSD since 386BSD 0.0new. Been around. And I always want to use FreeBSD on any servers, large and tiny, that I build, disk diskless, heady and headless, booting from disk, network, compact flash, whatever. I have done it. But with AWS there is a real bad problem and I am wondering why it doesn't get resolved for so many years? AWS after all is the most important hardware platform in today's world of servers. I have maintained a cage with HP DL585s and FibreChannel SAN, etc. but scrapped it all to move to "cloud" servers.
This here is a follow-up of Amazon AWS EC2 super light configuration a thread I started right here (with not a lot of feedback -- is this forum not the place to find real serious people? Are the serious people still preferring the email lists?)
Anyway on AWS there aresevere stability problems. just a few quirks that make it scary, such as:
appearing in the dmesg when booting a t3 instances.
When I boot a c5 or m5a it does not show up.
But worse is Even more scary, the system hung within hours, every time. That is, of course when we do more than what the average Linux guy can do with his Linux. In my case it is that I have reduced the system disk to only 1 GB by mounting the entire /usr filesystem via EFS (NFS). And lo and behold, after a few hours, say, 2-3 hours, the system freezes up, but there is no information at all that I can see. No kernel warning on the console screen shot, nothing in the log, no /var/log/messages, nothing. I have to force stop the instance and then restart it and all runs fine again, there is no sign in any logs of any trouble.
I have seen some stuff on the bugs reports and email lists, from 2018 https://lists.freebsd.org/pipermail/freebsd-current/2018-April/069140.html and https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=235856 but this seems to be unresolved. There is a weakness with HVM vs. paravirtualization, and with ENA network drivers. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=225791.
My trouble is, I have settled on this architecture with FreeBSD, and I just upgraded from 12.0 to 12.2 and nothing helped.
It is freezing up within around 4 hours.
The problem was I had not specified minorversion=1, it seemed to me that this was implied in oneopenown but it is necessary. And with that the hanging was gone.
-----------------
Hi, I've been with FreeBSD since 386BSD 0.0new. Been around. And I always want to use FreeBSD on any servers, large and tiny, that I build, disk diskless, heady and headless, booting from disk, network, compact flash, whatever. I have done it. But with AWS there is a real bad problem and I am wondering why it doesn't get resolved for so many years? AWS after all is the most important hardware platform in today's world of servers. I have maintained a cage with HP DL585s and FibreChannel SAN, etc. but scrapped it all to move to "cloud" servers.
This here is a follow-up of Amazon AWS EC2 super light configuration a thread I started right here (with not a lot of feedback -- is this forum not the place to find real serious people? Are the serious people still preferring the email lists?)
Anyway on AWS there are
Code:
intsmb0: <Intel PIIX4 SMBUS Interface> irq 20 at device 1.3 on pci0
intsmb0: Could not allocate I/O space
device_attach: intsmb0 attach returned 6appearing in the dmesg when booting a t3 instances.
When I boot a c5 or m5a it does not show up.
I have seen some stuff on the bugs reports and email lists, from 2018 https://lists.freebsd.org/pipermail/freebsd-current/2018-April/069140.html and https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=235856 but this seems to be unresolved. There is a weakness with HVM vs. paravirtualization, and with ENA network drivers. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=225791.
My trouble is, I have settled on this architecture with FreeBSD, and I just upgraded from 12.0 to 12.2 and nothing helped.
Code:
# Custom /etc/fstab for FreeBSD VM images
/dev/gpt/rootfs / ufs rw 1 1
/dev/gpt/varfs /var ufs rw 1 1
fdesc /dev/fd fdescfs rw 0 0
proc /proc procfs rw 0 0
/usr /.usr nullfs rw 0 0
fs-xxxxxx.efs.us-xxxxx-x.amazonaws.com:/ /usr nfs rw,nfsv4,minorversion=1,oneopenown,rsize=1048576,wsize=1048576,hard,timeo=600,retrans=2,noresvport,late,bg 0 0It is freezing up within around 4 hours.
The problem was I had not specified minorversion=1, it seemed to me that this was implied in oneopenown but it is necessary. And with that the hanging was gone.
