Hi,

 I am extremely sorry for being very late. I have reworked on the fix based
on the comments, I have adjusted the unwanted spaces reworked on the coding
style as suggested by Tirthankar. I have also incorporated the suggestions
made by Swathi.

The changes can be seen at
http://cr.opensolaris.org/~sdevulap/ha-dns_CR6530498/ . I would also like to
start a discussion on what changes need to be done in the documentation.

Regards,

Abhilash

2008/11/6 Abhilash T.G <abhi.tg4u at gmail.com>

> Hi,
>
>
>  I am right now busy with my end semester exam. I will look into all your
> suggestions as soon as my exams are over.
>
> Regards
> Abhilash
>
> 2008/11/6 Swathi Devulapalli <Swathi.Devulapalli at sun.com>
>
>  Hi Abhilash,
>>
>> Thanks for contributing to OHAC. Please find my comments below:
>>
>> 1. It would be good to have #def variables in capitals. aka ., #define
>> USER "user"
>>
>> 2. I doubt if this code will work as desired.
>>
>> *249                 if (errno == ERANGE) {*
>> *250                         ds_internal_error("Data buffer for user %s "
>> *
>> *251                                 "is not big enough",
>> user_name_st->val.val_str);*
>>
>> *252                         if (print_messages) {*
>> * 253                                 (void) fprintf(stderr,*
>> *254                                         gettext("Data buffer for
>> user %s "*
>> *255                                                 "is not big
>> enough.\n"),*
>> * 256
>> user_name_st->val.val_str);*
>> *257                                 }*
>> You are not returning here. Do you want to continue even with an error?
>> For this, you could move  return statement to line # 270. ie. after the
>> closing brace.
>>
>> *268                         return (1);*
>> *269                 }*
>> Where are you getting this "errno" from?
>>
>> 3. In the same code snip above, you are using ds_internal_error in the
>> "if" part and scds_syslog in the else part. Any specific reason?
>>
>> 4. */* Gets the username from the command line */*
>>
>> *562         strcpy(user_name, user_name_st->val.val_str);*
>>
>> *You may want to add (void) before this statement. Also, I feel it would
>> be better to do a strncpy with size as SCDS_ARRAY_SIZE.*
>>
>> *5. I do not see any debug statements in the code. You may want to
>> include some. *
>>
>> *6. Please check the indentations properly.*
>>
>> *7. Lastly, I would suggest you to do negative testing as well. Just
>> doing positive testing, may not help. Problems can be seen only in negative
>> testing.*
>>
>> *Thanks,*
>>
>> *Swathi*
>>
>>
>>
>> ----- Original Message -----
>> From: "Abhilash T.G" <abhi.tg4u at gmail.com>
>> Date: Wednesday, November 5, 2008 9:32 am
>> Subject: [ha-clusters-discuss] CR-6530498
>> To: clusters <ha-clusters-discuss at opensolaris.org>
>>
>> > Hello,
>> >
>> >   I am Abhilash T.G (OS0183). I have done the CR-6530498. The
>> > changes can
>> > be viewed at
>> > http://cr.opensolaris.org/~hari.sun/DNS_non-root_user_prop/webrev/<http://cr.opensolaris.org/%7Ehari.sun/DNS_non-root_user_prop/webrev/>
>> >
>> > The synopsis is ....
>> >
>> > *Bug ID*  6530498   *Synopsis*  HA DNS should provide a property
>> > to allow
>> > the DNS service to be run as a non-root user   *State*
>> > 1-Dispatched(Default State)
>> > *Category:Subcategory*  suncluster:ha-dns   *Keywords*  DNS |
>> > named |
>> > non-root | oss-bite-size | oss-request | user   *Sponsor*
>> >  *Submitter*
>> >  *Reported Against*
>> >  *Duplicate Of*
>> >  *Introduced In*
>> >  *Commit to Fix*
>> >  *Fixed In*
>> >  *Release Fixed*
>> >  *Related Bugs*
>> >  *Submit Date*  02-MAR-2007   *Last Update Date*  20-DEC-2007   *
>> > Description*
>> >
>> > Customer request via the Sun Cluster forum:
>> >
>> > I'd like the DNS HA Agent for Sun Cluster 3.2 to be modified so that
>> > the admin can choose to run BIND as a
>> > non-root user ie. named. This functionality is desirable for security
>> > reasons, as it gives the admin a layer of protection in
>> > the event that BIND is compromised. As a non-root process, it
>> > won't be
>> > capable of doing
>> > as much damage as a compromised root process could. This behaviour is
>> > similar to how Apache works, and is already
>> > available by specifying "-u <userid>" when starting BIND from the
>> > command line, or setting 'user'
>> > and 'group' in method_credential in the SMF manifest.
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> > *Work Around*
>> >
>> > N/A
>> >
>> >
>> >
>> > My Sponsor Harish has done the testing...
>> >
>> > I request you all to please review the changes.
>> >
>> > Regards
>> > Abhilash
>> >
>> > --
>> > Have the courage to follow your heart and intuitions, they somehow
>> > alreadyknow what you truly wants
>> >
>>
>
>
>
> --
> Have the courage to follow your heart and intuitions, they somehow already
> know what you truly wants
>



-- 
Have the courage to follow your heart and intuitions, they somehow already
know what you truly wants
-------------- next part --------------
An HTML attachment was scrubbed...
URL: 
<http://mail.opensolaris.org/pipermail/ha-clusters-discuss/attachments/20090608/15359137/attachment.html>

Reply via email to