You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@trafficserver.apache.org by "Zeller, Jan (ID)" <ja...@id.unibe.ch> on 2010/07/27 13:35:27 UTC
out of room in hostdb for round-robin DNS data
Dear list,
I get lots of "out of room in hostdb for round-robin DNS data" in my syslog. Should I increase the value as described in
http://trafficserver.apache.org/docs/v2/admin/files.htm#records.config.HostDB ?
The following sentence confuses me :
"Note: For values above 200000, you must increase the value of the proxy.config.hostdb.storage_size variable by at least 44 bytes per entry."
I can't find "proxy.config.hostdb.storage_size" in records.config
I am actually using 2.1.1-unstable & "./traffic_line -r proxy.process.hostdb.total_lookups" told me : 249351
thank you
---
Jan
Re: AW: out of room in hostdb for round-robin DNS data
Posted by Steve Jiang <sj...@apache.org>.
It would be helpful if you could narrow down the specific request that
is triggering the log message, or perhaps the log message occurs for
every request after a certain point? If it is for a specific request,
use the 'dig' tool to look at the DNS record for that hostname.
The hostdb is TS's DNS cache (TS does not use gethostbyname). The out
of room message indicates that the DNS lookup will not be cached, so the
proxy will do additional DNS queries in the future. This might impact
performance if your name server is not nearby.
--Steve
Zeller, Jan (ID) wrote, on 7/27/10 11:44 AM:
> Hi Steve,
>
>> Are you passing a lot of traffic with unique hostnames through TS or is
>> it an issue with some hostnames having a large number of DNS A records?
>
> to be honest I don't know what you mean by that.
>
> I have a cache-only bind which is listening on localhost installed on the same box.
> the resolv.conf looks like this :
>
> domain my.domain
> nameserver 127.0.0.1
> nameserver myprimarynameserver
> nameserver mysecondarynameserver
>
> Is there any relationship ? and what consequences does 'out of room...' finally have ?
>
>
>> Can you try adding the above line to your records.config and maybe doubling the size?
> I've pasted 'CONFIG proxy.config.hostdb.storage_size INT 67108864' into records.config and reloaded the config by applying 'traffic_line -x'
> Unfortunately the message persists.
>
> oops I restarted TS and
>
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: configuration changed: [hostdb.config] : reinitializing database
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: reconfiguring host database
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: Configured store too small, unable to reconfigure
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: unable to initialize database (too little storage) : [hostdb.config] : disabling database You may need to 'reconfigure' your cache manually. Please refer to the 'Configuration' chapter in the manual.
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: could not initialize host database. Host database will be disabled
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: bad hostdb or storage configuration, hostdb disabled
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: cache clustering disabled
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: logging initialized[7], logging_mode = 3
> Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: traffic server running
> Jul 27 20:23:33 traffic_server[19942]: {1080052048} NOTE: cache enabled
> Jul 27 20:23:44 traffic_cop[19718]: (http test) received non-200 status(502)
> Jul 27 20:23:44 traffic_cop[19718]: server heartbeat failed [1]
> .
> .
>
> That was because I set CONFIG proxy.config.hostdb.size INT 500000 before without having set CONFIG proxy.config.hostdb.storage_size
>
> Now I actually have
>
> CONFIG proxy.config.hostdb.size INT 200000
> CONFIG proxy.config.hostdb.storage_size INT 67108864
>
> these values seem to be alright despite the fact that I don't really understand that parametres.
>
> thanks,
>
> ---
>
> Jan
>
>
>
AW: out of room in hostdb for round-robin DNS data
Posted by "Zeller, Jan (ID)" <ja...@id.unibe.ch>.
Hi Steve,
>Are you passing a lot of traffic with unique hostnames through TS or is
>it an issue with some hostnames having a large number of DNS A records?
to be honest I don't know what you mean by that.
I have a cache-only bind which is listening on localhost installed on the same box.
the resolv.conf looks like this :
domain my.domain
nameserver 127.0.0.1
nameserver myprimarynameserver
nameserver mysecondarynameserver
Is there any relationship ? and what consequences does 'out of room...' finally have ?
>Can you try adding the above line to your records.config and maybe doubling the size?
I've pasted 'CONFIG proxy.config.hostdb.storage_size INT 67108864' into records.config and reloaded the config by applying 'traffic_line -x'
Unfortunately the message persists.
oops I restarted TS and
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: configuration changed: [hostdb.config] : reinitializing database
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: reconfiguring host database
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: Configured store too small, unable to reconfigure
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: unable to initialize database (too little storage) : [hostdb.config] : disabling database You may need to 'reconfigure' your cache manually. Please refer to the 'Configuration' chapter in the manual.
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: could not initialize host database. Host database will be disabled
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} WARNING: bad hostdb or storage configuration, hostdb disabled
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: cache clustering disabled
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: logging initialized[7], logging_mode = 3
Jul 27 20:23:33 traffic_server[19942]: {47887539635696} NOTE: traffic server running
Jul 27 20:23:33 traffic_server[19942]: {1080052048} NOTE: cache enabled
Jul 27 20:23:44 traffic_cop[19718]: (http test) received non-200 status(502)
Jul 27 20:23:44 traffic_cop[19718]: server heartbeat failed [1]
.
.
That was because I set CONFIG proxy.config.hostdb.size INT 500000 before without having set CONFIG proxy.config.hostdb.storage_size
Now I actually have
CONFIG proxy.config.hostdb.size INT 200000
CONFIG proxy.config.hostdb.storage_size INT 67108864
these values seem to be alright despite the fact that I don't really understand that parametres.
thanks,
---
Jan
Re: out of room in hostdb for round-robin DNS data
Posted by Steve Jiang <sj...@apache.org>.
Hi Jan,
Are you passing a lot of traffic with unique hostnames through TS or is
it an issue with some hostnames having a large number of DNS A records?
The proxy.config.hostdb.storage_size variable is not in the default
records.config file but it is configurable manually.
I haven't tested this specific variable in recent versions (especially
64bit). The default value is 32MB:
CONFIG proxy.config.hostdb.storage_size INT 33554432
Can you try adding the above line to your records.config and maybe
doubling the size?
--Steve
Zeller, Jan (ID) wrote, on 7/27/10 4:35 AM:
> Dear list,
>
> I get lots of "out of room in hostdb for round-robin DNS data" in my syslog. Should I increase the value as described in
>
> http://trafficserver.apache.org/docs/v2/admin/files.htm#records.config.HostDB ?
>
> The following sentence confuses me :
>
> "Note: For values above 200000, you must increase the value of the proxy.config.hostdb.storage_size variable by at least 44 bytes per entry."
>
> I can't find "proxy.config.hostdb.storage_size" in records.config
>
> I am actually using 2.1.1-unstable & "./traffic_line -r proxy.process.hostdb.total_lookups" told me : 249351
>
>
> thank you
>
> ---
>
> Jan
>