Mailing-List: contact derby-dev-help@db.apache.org; run by ezmlm
Precedence: bulk
Reply-To: <derby-dev@db.apache.org>
Message-ID: <5510566.1193734370932.JavaMail.jira@brutus>
Date: Tue, 30 Oct 2007 01:52:50 -0700 (PDT)
From: =?utf-8?Q?J=C3=B8rgen_L=C3=B8land_=28JIRA=29?= <jira@apache.org>
To: derby-dev@db.apache.org
Subject: [jira] Commented: (DERBY-3064) Implement the LogShipper that will
 enable the shipping of Log records from the master to the slave
In-Reply-To: <13994899.1189420769956.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable


    [ https://issues.apache.org/jira/browse/DERBY-3064?page=3Dcom.atlassian=
.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12538704 ]=20

J=C3=B8rgen L=C3=B8land commented on DERBY-3064:
--------------------------------------

> (snip) And exceptions are automatically logged in accordance with derby.s=
tream.error.logSeverityLevel anyway, aren't they? If the user has requested=
 no error logging, I'm not sure if we should ignore that request. (If the u=
ser is never going to see the exception, I think it's a good idea to log it=
, but in startMaster() the exception will propagate out to the user, if I u=
nderstand correctly.)

Knut,
You are right about startMaster - exceptions thrown from this method would =
propagate back to the user. Apart from that single method, however, the met=
hods in MasterController are called by either

1. a user transaction thread executing normal transactions, or
2. the checkpoint daemon, or
3. the log shipper thread

Thus, we need a plan for what to do with exceptions from these methods. Eve=
n though it would be possible to propagate a (wrapped) ReplicationLogBuffer=
Full exception back to a random user who is executing a transaction (scenar=
io 1.), I'm not sure that's the right address for these messages. That's wh=
y exceptions are currently not thrown to the callers of the MasterFactory (=
interface of MC) methods.

Would it suffice to use MC#printErrorStack for exceptions in all methods ex=
cept startMaster, and let MC#printErrorStack print error messages based on =
the value of a property, e.g. "derby.replication.verbose"? The property cou=
ld, e.g., correspond to the same logSeverityLevel used for other exceptions=
. That way, the amount of information written to derby.log can be controlle=
d.

If this sounds like a good idea, I will add the aforementioned property in =
a SlaveController patch that will make use of the functionality added in De=
rby-3071.

> Implement the LogShipper that will enable the shipping of Log records fro=
m the master to the slave
> -------------------------------------------------------------------------=
-------------------------
>
>                 Key: DERBY-3064
>                 URL: https://issues.apache.org/jira/browse/DERBY-3064
>             Project: Derby
>          Issue Type: Sub-task
>          Components: Miscellaneous
>            Reporter: V.Narayanan
>            Assignee: V.Narayanan
>         Attachments: LogShipperImpl_v1.diff, LogShipperImpl_v1.stat, LogS=
hipperImpl_v2.diff, LogShipperImpl_v2.stat, LogShipperImpl_v3.diff, LogShip=
perImpl_v3.stat, LogShipperImpl_v4.diff, LogShipperImpl_v4.stat, LogShipper=
Impl_v5.diff, LogShipperImpl_v5.stat, LogShipperIntegration_v1.diff, LogShi=
pperIntegration_v1.stat, LogShipperIntegration_v2.diff, LogShipperIntegrati=
on_v2.stat
>
>


--=20
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.