I have begun to notice that the smartd stalls my machine's boot process for as much as 2-3 minutes while it is coming up. I have the BIOS SMART function enabled, and I removed the bad SG400GB drive from the box yesterday, but I think this goes back farther than that.
Why does it take so long? Is the BIOS SMART involved in this (and how)?
Also, sendmail and sm-client take 30s-2m to come up as well (not always, but usually).
Config: CentOS 4.4 Plus (2.6.9-42.0.10 kernel) AMD Athlon 64 x2 4200+ ECS NFORCE4M-A m/b 2GB OCZ DDR2 800 (PC6400) ram 160MB + 120MB Maxtor PATA drives (160 is the boot drive) 300GB Seagate SATA drive (/boot and / are here) geFORCE7100sg (nVidia) PCI-E x16 video card Minolta QMS PagePro 1100 laser printer on /dev/lp0
Thanks.
Mark Hull-Richter, Linux Kernel Engineer DATAllegro (www.datallegro.com) 85 Enterprise, Second Floor, Aliso Viejo, CA 92656 949-680-3082 - Office 949-330-7691 - fax
Mark Hull-Richter spake the following on 4/18/2007 10:59 AM:
I have begun to notice that the smartd stalls my machine's boot process for as much as 2-3 minutes while it is coming up. I have the BIOS SMART function enabled, and I removed the bad SG400GB drive from the box yesterday, but I think this goes back farther than that.
Why does it take so long? Is the BIOS SMART involved in this (and how)?
The smart bios option is usually only to give you an error message on boot. Sendmail start delays are usually related to name resolution, and can sometimes be related to entries in /etc/hosts. Do you have an sata controller that isn't supported with smartd?
On 4/18/07, Scott Silva ssilva@sgvwater.com wrote:
The smart bios option is usually only to give you an error message on
boot.
Sendmail start delays are usually related to name resolution, and can sometimes be related to entries in /etc/hosts.
Here's mine:
# Do not remove the following line, or various programs # that require network functionality will fail. 69.234.27.32 mhrichter 127.0.0.1 mhrichter localhost.localdomain localhost 192.168.108.128 mhr-wm-win
Do you have an sata controller that isn't supported with smartd?
I don't think so - it's the on-board controller in the mobo (nVidia chipset), and this only started when I started having problems with the SG 400GB drive (that is now gone), or at least that was when I began to notice it. My lsmod shows sata_nv, which would be the right one.
mhr
On 4/18/07, Mark Hull-Richter mhullrich@gmail.com wrote:
Here's mine:
# Do not remove the following line, or various programs # that require network functionality will fail. 69.234.27.32 mhrichter 127.0.0.1 mhrichter localhost.localdomain localhost 192.168.108.128 mhr-wm-win
It would really best to supply the whole host.domain in the file for the entry. Since you also have 2 different networks, sendmail may be having trouble figuring out what the server really is. The 192.168 address resolves differently than the 69.234 address. And the 69.234 address doesn't reverse to the same name set in /etc/hosts. This confusion can very much cause sendmail to get all turned around and stall. You might want to consider looking at your logs to see if you have errors reflecting this.
Do you have an sata controller that isn't supported with smartd?
I don't think so - it's the on-board controller in the mobo (nVidia chipset), and this only started when I started having problems with the SG 400GB drive (that is now gone), or at least that was when I began to notice it. My lsmod shows sata_nv, which would be the right one.
Are you using the stock drivers or the proprietary nForce chipset drivers?
On 4/18/07, Jim Perrin jperrin@gmail.com wrote:
It would really best to supply the whole host.domain in the file for the entry. Since you also have 2 different networks, sendmail may be having trouble figuring out what the server really is. The 192.168 address resolves differently than the 69.234 address. And the 69.234 address doesn't reverse to the same name set in /etc/hosts. This confusion can very much cause sendmail to get all turned around and stall. You might want to consider looking at your logs to see if you have errors reflecting this.
Looks like it - I keep seeing:
Apr 18 09:07:45 mhrichter sendmail[3376]: My unqualified host name (mhrichter) unknown; sleeping for retry Apr 18 09:08:45 mhrichter sendmail[3376]: unable to qualify my own domain name (mhrichter) -- using short name
The original host name was something like adsl-69-234-xx-xx-<something>.<an att thing>.com. IIUTC, you're saying I should include that original domain name on the host name line.
The 192.138 address is a local subdomain for the VMWare server NAT (which works really nicely) - would I need to modify that one, too or just the host?
Are you using the stock drivers or the proprietary nForce chipset drivers?
Whatever CentOS used when it came up from the install, most likely the stock drivers. I haven't looked to see if there are any updates since these were working fine from install to present. I'll see what's there....
Thanks.
On 4/18/07, Mark Hull-Richter mhullrich@gmail.com wrote:
On 4/18/07, Jim Perrin jperrin@gmail.com wrote:
It would really best to supply the whole host.domain in the file for the entry. Since you also have 2 different networks, sendmail may be having trouble figuring out what the server really is. The 192.168 address resolves differently than the 69.234 address. And the 69.234 address doesn't reverse to the same name set in /etc/hosts. This confusion can very much cause sendmail to get all turned around and stall. You might want to consider looking at your logs to see if you have errors reflecting this.
I did both this and took Scott's advice (to leave my machine name off the localhost line) and the sendmail and sm-client now start just fine (fast).
Are you using the stock drivers or the proprietary nForce chipset
drivers?
I couldn't find any sata drivers from nVidia, just video drivers.
However, I did notice that when I put a new sata drive in to replace the defective one, smartd was happy and quick once again. I thought maybe there was a bad entry in the smartd.conf file, but that drive isn't in there at all (I'll be adding it once it passes a preliminary badblocks screen - so far so good).
The new drive is a WD320 - a little smaller (and cheaper), but it seems to be running quite a bit better already....
Mark Hull-Richter wrote:
On 4/18/07, Mark Hull-Richter mhullrich@gmail.com wrote:
On 4/18/07, Jim Perrin jperrin@gmail.com wrote:
Are you using the stock drivers or the proprietary nForce chipset
drivers?
I couldn't find any sata drivers from nVidia, just video drivers.
http://www.nvidia.com/object/linux_nforce_1.21.html - that wasn't too hard ...
Although the sata_nv driver is available as a module in the stock CentOS 4 kernel.
Cheers,
Ralph
Mark Hull-Richter spake the following on 4/18/2007 11:27 AM:
On 4/18/07, *Scott Silva* <ssilva@sgvwater.com mailto:ssilva@sgvwater.com> wrote:
> The smart bios option is usually only to give you an error message on boot. > Sendmail start delays are usually related to name resolution, and can > sometimes be related to entries in /etc/hosts.Here's mine:
# Do not remove the following line, or various programs # that require network functionality will fail. 69.234.27.32 http://69.234.27.32 mhrichter
Do you have a FQDN for this one? Might not be necessary, but worth a shot.
127.0.0.1 http://127.0.0.1 mhrichter
And just have localhost.localdomain localhost here
localhost.localdomain localhost 192.168.108.128 http://192.168.108.128 mhr-wm-win
> Do you have an sata controller that isn't supported with smartd?I don't think so - it's the on-board controller in the mobo (nVidia chipset), and this only started when I started having problems with the SG 400GB drive (that is now gone), or at least that was when I began to notice it. My lsmod shows sata_nv, which would be the right one.
mhr
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
-
Jim Perrin -
Mark Hull-Richter -
Ralph Angenendt -
Scott Silva