Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
HBase >> mail # dev >> Cannot locate root region


Copy link to this message
-
RE: Cannot locate root region
Yup totally - either name or ip would work. Not sure if there is a pro or a con to choosing either one - but thought it better to use the ip as that always remains the same (no resolve required) and used to open the sockets.

@jd-cryans: Saw your JIRA update: "One example of weirdness is when the region server is told which address to use according to the master:"

Was meaning to ask about that too :) its all good now.

Thanks
Karthik
-----Original Message-----
From: Joydeep Sarma [mailto:[EMAIL PROTECTED]]
Sent: Friday, January 29, 2010 11:01 AM
To: [EMAIL PROTECTED]
Subject: Re: Cannot locate root region

hmmm .. if the master doesn't need the RS ip address at this point -
seems like it should be able to use the hostname offered by the RS
directly?

On Fri, Jan 29, 2010 at 10:44 AM, Karthik Ranganathan
<[EMAIL PROTECTED]> wrote:
> The master does another lookup independent of the region server using the hostname given by the region server:
>
> ServerManager.java, regionServerReport() does:
>    HServerInfo storedInfo = serversToServerInfo.get(info.getServerName()); // info.getServerName() is hostname
>
> Which eventually does:
>        HServerAddress.getHostname()
>
> HServerAddress' constructor creates the InetSocketAddress from the hostname:port, which involves mapping the hostname to the ip address using a lookup.
>
> Thanks
> Karthik
>
>
> -----Original Message-----
> From: Joydeep Sarma [mailto:[EMAIL PROTECTED]]
> Sent: Friday, January 29, 2010 9:46 AM
> To: [EMAIL PROTECTED]
> Subject: Re: Cannot locate root region
>
> @Kannan - Karthik's mail said the reverse lookup happens in the RS
> (not the master). the master simply tried to match the offered
> hostname.
>
> i dont know whose reading is right - but if it's the RS - i didn't
> understand why that wasn't just the reverse lookup done once at
> bootstrap time (which wouldn't be affected by ongoing DNS badness).
>
>
> On Fri, Jan 29, 2010 at 9:39 AM, Jean-Daniel Cryans <[EMAIL PROTECTED]> wrote:
>> I just created https://issues.apache.org/jira/browse/HBASE-2174
>>
>> We handle addresses in different ways depending on which part of the
>> code you're in. We should correct that everywhere by implementing a
>> solution that also solves what you guys are seeing.
>>
>> J-D
>>
>> On Fri, Jan 29, 2010 at 8:33 AM, Kannan Muthukkaruppan
>> <[EMAIL PROTECTED]> wrote:
>>> @Joy: The info stored in .META. for various regions as well as in the ephemeral nodes for region servers in zookeeper are both already IP address based. So doesn't look like multi-homing and/or the other flexibilities you mention were a design goal as far as I can tell.
>>>
>>> Regarding: <<< doesn't the reverse ip lookup just once at RS startup time?>>>, what seems to be happening is this:
>>>
>>> A regionServer periodically sends a regionServerReport (RPC call) to the master. A HServerInfo argument is passed as an argument and it identifies the sending region server's identity in IP address format.
>>>
>>> The master, in ServerManager class, maintains a serversToServerInfo map which is hostname based. Every time a master receives a regionServerReport it converts the IP address based name to a hostname via the info.getServerName() call. Normally this call returns the hostname, but we suspect that during the DNS flakiness, it returned an IP address based string. And so, this caused ServerManager.java to think that it was hearing from a new server. And this lead to:
>>>
>>>  HServerInfo storedInfo = serversToServerInfo.get(info.getServerName());
>>>    if (storedInfo == null) {
>>>      if (LOG.isDebugEnabled()) {
>>>        LOG.debug("Received report from unknown server -- telling it " +   <<===========>>>          "to " + CALL_SERVER_STARTUP + ": " + info.getServerName());  <<===========>>>      }
>>>
>>> and bad things down the road.
>>>
>>> The above error message in our logs (example below) indeed identified the host in IP address syntax, even though normally the getServerName call would return the info in hostname format.