password hash and db link is a risk (was OEM permissions)

2003-12-24 Thread Pete Finnigan
Hi There are risks with knowing the password hash for any user not just system. Unfortunately reading it from dba_users is not the only way to get it. There are a number of other options as well that i won't go into here. As Arup pointed out the password hash is calculated from the username and

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Rachel Carmichael
9.2.0.2 on Sun Solaris... and yes, I got the same encrypted password --- Michael Thomas [EMAIL PROTECTED] wrote: Hi, Okay. I'm almost a believer of this as a problem. How about 9.2.0.4 on RH9.3. 1) What does anyone/everyone get for my this query (my results shown): connect

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Scott Canaan
This is what I got, Oracle 8.1.7.4 on Sun Solaris (I dropped the user): Connected to: Oracle8i Enterprise Edition Release 8.1.7.4.0 - Production With the Partitioning option JServer Release 8.1.7.4.0 - Production SQL create user scott identified by tiger; User created. SQL select password 2

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Jamadagni, Rajendra
[EMAIL PROTECTED] . oraenv ORACLE_SID = [OLDNCS1] ? DEVL [EMAIL PROTECTED] sys SQL*Plus: Release 9.2.0.2.0 - Production on Tue Dec 23 08:30:45 2003 Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved. Connected to: Oracle9i Enterprise Edition Release 9.2.0.2.0 - 64bit Production

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Norris, Gregory T [ITS]
Same here... SQL select password from dba_users where username = 'SCOTT'; PASSWORD -- F894844C34402B67 -Original Message- Michael Thomas Sent: Monday, December 22, 2003 10:44 PM To: Multiple recipients of list ORACLE-L Hi, Okay. I'm almost a believer of

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Stephen.Lee
When I brought the issue up, I didn't know if one could in fact maliciously use that info. And, as I originally stated, it was something I had not tried. But paranoia (healthy, I think) dictates there's gotta be a way. When one looks at the Unix password world which brought about the necessity

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Jared Still
As long Oracle can manage to keep its modified DES algorithm secret, this should make it somewhat difficult to crack passwords in the manner that can be done with unix passwords. It could still be done, but the time required would make it just too time consuming IMO. Jared On Tue, 2003-12-23

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Norris, Gregory T [ITS]
Not really... you could easily compile a list of passwords and their associated hashes. Once this is done, it's just a simple matter of matching the hashes. -Original Message- Jared Still Sent: Tuesday, December 23, 2003 1:24 PM To: Multiple recipients of list ORACLE-L As long Oracle

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Davey, Alan
No. Two different users with the same password would have different hash values. So you would have to loop through a dictionary list for each user within your local database. Once you got a match, then you could logon to the target database with that user/password combo.

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Norris, Gregory T [ITS]
Ah, you're right... the username is taken into account when the password is hashed. Quite a few applications have forced usernames, however, so I'm still a bit uneasy. It's clearly not as bad as I thought, however. -Original Message- Davey, Alan Sent: Tuesday, December 23, 2003 2:59

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Arup Nanda
Actually, the concatenated string of userid and password is hashed. So if that is same, you got yourself the same hashed password. Consider this: SQL create user ABC identified by DEF; User created. SQL create user ABCD identified by EF; User created. SQL select password from dba_users where

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Jared Still
You could conceivably do this, much like lopht or crack. It would take a rather large password database, and a cracker with some intelligence. This is the same reason that unix now uses shadow passwords. Jared On Tue, 2003-12-23 at 12:29, Norris, Gregory T [ITS] wrote: Not really... you

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Davey, Alan
Very Nice. I didn't know how the 2 values were used within the hashing algorithm. I would have thought it was a little more complex. - Alan Davey Senior Analyst/Project Leader Oracle 9i OCA; 3/4 OCP w) 973.267.5990 x458 w) 212.295.3458 -Original

RE: Risk of knowing password hash value (Was: OEM permissions)

2003-12-23 Thread Bellow, Bambi
Looks like they're using VMS's algorithm. *That's* a shocker! -Original Message- Sent: Tuesday, December 23, 2003 3:45 PM To: Multiple recipients of list ORACLE-L Actually, the concatenated string of userid and password is hashed. So if that is same, you got yourself the same hashed

Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread Yong Huang
Hi, Gregory, I only have access to Oracle 9.2 on my laptop. Here's my test. I have ORCL and AUX1 databases, the latter created by RMAN DUPLICATE some time ago. I logon AUX1 as SYSTEM. Set SYSTEM password hash value to the same as in ORCL. Create link L to ORCL without password. Selecting from a

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread Jared Still
Environment: DB1: RH 8.0 with Oracle EE 9.2.0.4 DB2: Win2k SP3 with Oracle EE 9.2.0.1 SYSTEM user on each database initially have different passwords. It goes something like this: DB1: select password from dba_users where username = 'SYSTEM'; Let's say the result is 'AC424SDK4398' DB2:

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread Yong Huang
Jared, I see you log out and log back in as SYSTEM to DB2. But how do you know the password for SYSTEM to log back in with after you change it? What if you don't log out? When I tried that (i.e. not logging out), I got ORA-1017. Yong Huang --- Jared Still [EMAIL PROTECTED] wrote: Environment:

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread Michael Thomas
Hi, Okay. I'm almost a believer of this as a problem. How about 9.2.0.4 on RH9.3. 1) What does anyone/everyone get for my this query (my results shown): connect system/[EMAIL PROTECTED]; alter user scott identified by tiger; -- select password from dba_users where username = 'SCOTT'; PASSWORD

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread Jared Still
It doesn't matter which account I logged into DB2 with, as long as that account has privileges to read DBA_USERS. SYSTEM was used simply because it was the only account on the database that could be logged into remotely, so my test could be run without switching between machines. If I had

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread Jared Still
On RH 8.0 Oracle 9.2.0.4: F894844C34402B67 It is required that a password for a particular users always hashes to the same value, regardless of platform or Oracle version. This has been true for as long as I have used oracle: since 7.0.13. If not, export/import would not be able to recreate

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread Jared Still
Yes, I misunderstood. Once I change the password, I can no longer connect to the account. My hasty little test was missing an important condition: I should have pretended I didn't know the password to the other database, which would prevent me from logging back on exploiting the db link. Wonder

Re: Risk of knowing password hash value (Was: OEM permissions)

2003-12-22 Thread rahul sharma
8.1.7 on win2000 SQL select password 2 from dba_users 3 where username = 'SCOTT'; PASSWORD -- F894844C34402B67 - Original Message - To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED] Sent: Tuesday, December 23, 2003 11:44 AM Hi, Okay.

RE: OEM permissions

2003-12-20 Thread Yong Huang
Hi, I think you're describing a real security hole. But I'm not sure how it's exploited exactly. Let's say John Doe sets up his database on his desktop, which is part of the production database network. He sees the hash value of SYSTEM's password on production and sets the hash value for his own

RE: OEM permissions

2003-12-20 Thread Jared Still
You could use protocol.ora to specify which machines are allowed to make a connection to the database. In some environments this would be fairly painless. SAP for example. The users do not connect to the database, they connect to the app server. The number of machines that legitimately require

RE: OEM permissions

2003-12-20 Thread Thomas A. La Porte
I believe the point is not that you can create links to SYS or SYSTEM accounts, but instead to application accounts, e.g. if I created a link from my private database to the company's HR database using a duplicated HR_MANAGER schema, I may be able to access data that I otherwise should not

RE: OEM permissions

2003-12-20 Thread Norris, Gregory T [ITS]
There's no reason I can see that he couldn't create the dblink first, and then reset the password using the encrypted value. Alternately, the dblink could be created using the DBMS_SYS_SQL package... no knowledge of the current password required. create database link foo

RE: OEM permissions

2003-12-19 Thread jo_holvoet
-L [EMAIL PROTECTED] cc: Subject:RE: OEM permissions Raj - Thanks for your reply. Were this a consultant, my reply would mirror yours, and maybe not so diplomatically. But basically I manage these databases on behalf of this manager, so when he asks for read-only

RE: OEM permissions

2003-12-19 Thread Yong Huang
Hi, Raj, 9i doesn't allow a user with select any table privilege to view any object owned by SYS. So the sys.link$ risk is gone. But select any dictionary, a new privilege in 9i, allows that. In practice, I always grant select_catalog_role to any developer, but refrain from granting select any

RE: OEM permissions

2003-12-19 Thread Michael Thomas
A possibly related question: I'm curious if everyone allows your developers to see V$SQL... views? If not, then ... whatever ... no comment. I'm disappointed with some perspectives in these threads regarding developers. Rather than close doors, why not use 'development' instances, and role based

RE: OEM permissions

2003-12-19 Thread Stephen.Lee
recipients of list ORACLE-L Subject: RE: OEM permissions A possibly related question: I'm curious if everyone allows your developers to see V$SQL... views? If not, then ... whatever ... no comment. I'm disappointed with some perspectives in these threads regarding developers. Rather

RE: OEM permissions

2003-12-19 Thread Jesse, Rich
Sure it does, just not by default. Check out the 07_DICTIONARY_ACCESSIBILITY parameter. Enjoy! Rich Rich Jesse System/Database Administrator [EMAIL PROTECTED] Quad/Tech Inc, Sussex, WI USA -Original Message- Sent: Friday, December 19, 2003

RE: OEM permissions

2003-12-19 Thread Jamadagni, Rajendra
That's why we have scripts which give us a report every few days on users that have db_links, any of the *_ANY_* privs (like alter any table), dba privs etc, also a list of some sensitive schemas too. Our application support needed to work with users to grant/revoke roles and private synonyms.

OEM permissions

2003-12-18 Thread DENNIS WILLIAMS
We have a new manager that wants his group to use OEM for development access, as an alternative to Toad. He has requested a special Oracle userid with the following grants: SELECT_CATALOG_ROLE SELECT ANY DICTIONARY SELECT ANY TABLE Does this seem reasonable for OEM? The manager is

RE: OEM permissions

2003-12-18 Thread Jamadagni, Rajendra
Dennis, select any table has to be a big no no ... anyone can select from sys.link$. But I am still trying how OEM can be used for _development_?? what am I missing? As for One of our groups hired a new consultant and he (claimed to have DBA background) immediately shot off an email saying he

RE: OEM permissions

2003-12-18 Thread Stephen.Lee
Wouldn't this allow viewing DBA_USERS? I haven't tried this myself, but it seems that I could set up another oracle instance, create a user identified by values, then create database link. -Original Message- We have a new manager that wants his group to use OEM for development

RE: OEM permissions

2003-12-18 Thread DENNIS WILLIAMS
Raj - Thanks for your reply. Were this a consultant, my reply would mirror yours, and maybe not so diplomatically. But basically I manage these databases on behalf of this manager, so when he asks for read-only access, I can't really refuse. And I think he is pretty competent as a DBA. He says

RE: OEM permissions

2003-12-18 Thread Mercadante, Thomas F
Dennis, I think you are probably ok with this. But the best way to do this is to create an Oracle account, grant what he asks, and start OEM using that account. Try and change things and see what happens. Then you will know for sure what the impact is. Good Luck! Tom Mercadante Oracle

RE: OEM permissions

2003-12-18 Thread Odland, Brad
I would say OEM for dbas only. To difficult to manage for developers. Dangerous. Too much low level stuff for managers and developers to poke around with. OEM is not really a development tool. 2 cents -Original Message- Sent: Thursday, December 18, 2003 8:24 AM To: Multiple

RE: OEM permissions - thanks!

2003-12-18 Thread DENNIS WILLIAMS
Raj, Thomas, Brad, Bill, anyone I forgot - I feel better about the situation now. Thanks for letting me run this issue past you. Dennis Williams DBA Lifetouch, Inc. [EMAIL PROTECTED] -Original Message- Sent: Thursday, December 18, 2003 9:35 AM To: Multiple recipients of list ORACLE-L