Oracle date compare broken because of DST - oracle
We've been debugging an issue with a SQL query executed from an app server running Java via Hibernate. The error:
[3/10/14 10:52:07:143 EDT] 0000a984 JDBCException W org.hibernate.util.JDBCExceptionReporter logExceptions SQL Error: 1878, SQLState: 22008
[3/10/14 10:52:07:144 EDT] 0000a984 JDBCException E org.hibernate.util.JDBCExceptionReporter logExceptions ORA-01878: specified field not found in datetime or interval
We've been able to narrow this down to the simple SQL below.
select *
from MY_TABLE T
where T.MY_TIMESTAMP >= (CURRENT_TIMESTAMP - interval '1' hour );
When we run this in the same database, we get the error:
ORA-01878: specified field not found in datetime or interval
01878. 00000 - "specified field not found in datetime or interval"
*Cause: The specified field was not found in the datetime or interval.
*Action: Make sure that the specified field is in the datetime or interval.
The MY_TIMESTAMP column is defined as TIMESTAMP(6).
FWIW, if we change the comparison in the SQL above from >= to <=, the query works.
We assume this has something to do with the time change (we're in America/New_York) but we're having problems trying to figure out where to go from here with our debugging.
Also, we've seen this problem with a similar query that's running through MyBatis and the error looks like:
### Error querying database. Cause: java.sql.SQLException: ORA-01878: specified field not found in datetime or interval
### The error may involve defaultParameterMap
### The error occurred while setting parameters
### Cause: java.sql.SQLException: ORA-01878: specified field not found in datetime or interval
UPDATE: A teammate on Windows changed her Windows Date and Time settings by un-checking "Automatically adjust clock for Daylight Saving Time" and then opened a new SQLDeveloper instance. The second instance is able to run the query without any issue but the first (with the old DST setting) still fails.
To avoid this error, consider using an explicit cast of the expression in the where clause to a timestamp type (timestamp without timezone), in this way:
select *
from MY_TABLE T
where T.MY_TIMESTAMP >= cast(CURRENT_TIMESTAMP - interval '1' hour As timestamp );
Alternatively you can explicitely set the session time zone to, for example '-05:00' - for New York standard (winter) time, using ALTER SESSION time_zone = '-05:00', or by setting ORA_SDTZ environment variable in all client's environments, see this link for details: http://docs.oracle.com/cd/E11882_01/server.112/e10729/ch4datetime.htm#NLSPG263
But it also depends on what really is stored in the timestamp column in the table, for example what a timestamp 2014-07-01 15:00:00 represents in fact, is it a "winter time" or a "summer time" ?
CURRENT_TIMESTAMP function returns a value of datatype TIMESTAMP WITH TIME ZONE
see this link: http://docs.oracle.com/cd/B19306_01/server.102/b14200/functions037.htm
While comparing timestamps and dates, Oracle implicitely converts the data to the more precise data type using the session time zone !
See this link --> http://docs.oracle.com/cd/E11882_01/server.112/e10729/ch4datetime.htm#NLSPG251
In our particular case, Oracle casts timestamp column to the timestamp with time zone type.
Oracle determines a session timezone from the client environment.
You can determine current session timezone using this query:
select sessiontimezone from dual;
For example on my PC (Win 7), when the option ""Automatically adjust clock for Daylight Saving Time" is checked, this query returns (under SQLDeveloper):
SESSIONTIMEZONE
---------------
Europe/Belgrade
When i uncheck this option in Windows and then restart SQLDeveloper, it gives:
SESSIONTIMEZONE
---------------
+01:00
The former session timezone is a timezone with a region name, for which Oracle uses the Daylight Saving Time rules for this region in date calculations:
alter session set time_zone = 'Europe/Belgrade';
select cast( timestamp '2014-01-29 01:30:00' as timestamp with time zone ) As x,
cast( timestamp '2014-05-29 01:30:00' as timestamp with time zone ) As y
from dual;
session SET altered.
X Y
---------------------------- ----------------------------
2014-01-29 01:30:00 EUROPE/B 2014-05-29 01:30:00 EUROPE/B
ELGRADE ELGRADE
The latter timezone uses a fixed offset "+01:00" (always the "Winter time"), and Oracle does not apply any DST rules for it, it simply adds the fixed offset.
alter session set time_zone = '+01:00';
select cast( timestamp '2014-01-29 01:30:00' as timestamp with time zone ) As x,
cast( timestamp '2014-05-29 01:30:00' as timestamp with time zone ) As y
from dual;
session SET altered.
X Y
---------------------------- ----------------------------
2014-01-29 01:30:00 +01:00 2014-05-29 01:30:00 +01:00
Please note, for curiosity's sake, that Y results in the above represent two different times !!!
014-05-29 01:30:00 EUROPE/BELGRADE is not the same as: 2014-05-29 01:30:00 +01:00
but actually this:
014-05-29 01:30:00 EUROPE/BELGRADE is equal to: 2014-05-29 01:30:00 +02:00
The above is only to make you aware of how simple "box un-checking" could affect your queries, and where to dig for a reason when users complain "this query worked fine in January, but gave wrong results in July".
And still on the topic of ORA-01878 - let say my session is EUROPE/Warsaw and my table containts this timestamp (without time zone)
'TIMESTAMP'2014-03-30 2:30:00'
Note that in my region the DST change, in 2014 year, occurs on 30 of march at 2:00 a.m.
It simply means that on march 30, at 2:00 at night, I must wake up and shift my watch forward from 2:00 to 3:00 ;)
alter session set time_zone = 'Europe/Warsaw';
select cast( TIMESTAMP'2014-03-30 2:30:00' as timestamp with time zone ) As x
from dual;
SQL Error: ORA-01878: podane pole nie zostaĆo znalezione w dacie-godzinie ani w interwale
01878. 00000 - "specified field not found in datetime or interval"
*Cause: The specified field was not found in the datetime or interval.
*Action: Make sure that the specified field is in the datetime or interval.
Oracle knows, that this timestamp is not valid in my region according to DST rules, because there is no time 2:30 on 30 of march - at 2:00 the clock is moved to 3:00, and there is no time 2:30. Therefore Oracle throws the error ORA-01878.
However this query works perfectly fine:
alter session set time_zone = '+01:00';
select cast( TIMESTAMP'2014-03-30 2:30:00' as timestamp with time zone ) As x
from dual;
session SET altered.
X
----------------------------
2014-03-30 02:30:00 +01:00
And this is a reason of this error - your table contains timestamps like 2014-03-09 2:30 or so (for New York, where DST shifts occur on 9 of March and 2 of November), and Oracle doesn't know how to convert them from timestamp (without TZ) to timestamp with TZ.
The last question - why the query with >= doesn't work, but the query with <= works fine ?
They work/don'n work, because SQLDeveloper returns only first 50 rows (maybe 100 ? It depends on settings). The query doesn't read the whole table, it stops when first 50(100) rows are fetched.
Change the "working" query to, for example:
select sum( EXTRACT(HOUR from MY_TIMESTAMP) ) from MY_TABLE
where MY_TIMESTAMP <= (CURRENT_TIMESTAMP - interval '1' hour );
This force the query to read all rows in the table, and the error will appear, I'am 100% sure.
Thanks to kordirko for the extremely detailed write up. I think in the future, we will be looking at different ways to compare dates that aren't as prone to error. In the meantime, we were able to figure out the problem and both a temporary and long-term solution.
First, more details on the issue. It turns out that the values being stored in the TIMESTAMP field in the database were incorrect. We saw this by using the dump function and examining the bytes. If you look at the 5th byte in the output below, you'll see the hour (this is actually the hour + 1 so 5 is actually 4AM). For the values between 3AM and 4AM, you'll notice that the 5th byte is 3 which represents 2AM. 2 AM March 9, 2014 in EST doesn't exist - this is an incorrect time according to DST rules and Oracle's rules.
09-MAR-14 03.06.21.522000000 AM Typ=180 Len=11: 120,114,3,9,3,7,22,31,29,22,128
09-MAR-14 03.32.37.869000000 AM Typ=180 Len=11: 120,114,3,9,3,33,38,51,203,227,64
09-MAR-14 03.36.49.804000000 AM Typ=180 Len=11: 120,114,3,9,3,37,50,47,236,17,0
09-MAR-14 03.43.47.328000000 AM Typ=180 Len=11: 120,114,3,9,3,44,48,19,140,226,0
09-MAR-14 03.47.55.255000000 AM Typ=180 Len=11: 120,114,3,9,3,48,56,15,50,253,192
09-MAR-14 03.55.45.129000000 AM Typ=180 Len=11: 120,114,3,9,3,56,46,7,176,98,64
09-MAR-14 04.05.03.325000000 AM Typ=180 Len=11: 120,114,3,9,5,6,4,19,95,27,64
09-MAR-14 04.28.41.267000000 AM Typ=180 Len=11: 120,114,3,9,5,29,42,15,234,24,192
09-MAR-14 04.35.16.072000000 AM Typ=180 Len=11: 120,114,3,9,5,36,17,4,74,162,0
09-MAR-14 04.41.07.260000000 AM Typ=180 Len=11: 120,114,3,9,5,42,8,15,127,73,0
09-MAR-14 04.46.31.047000000 AM Typ=180 Len=11: 120,114,3,9,5,47,32,2,205,41,192
09-MAR-14 04.53.33.471000000 AM Typ=180 Len=11: 120,114,3,9,5,54,34,28,18,227,192
After much research and discussion, we zeroed in on the fact that our version of the Oracle JDBC driver (11.2.0.2) might've been inserting the bad values. Oracle's information page on 11.2.0.3 references a bug that looks like it's our issue: "9785135 DST conversion not correct using jdbc 11g timestamtz". We wrote a quick test class that inserts values from March 9, 2014 1:50 AM to 4:00 AM using both the 11.2.0.2 and 11.2.0.3 driver. Here's a snippet of what was inserted into the db:
DRIVER_V JAVA_DATE_AS_STRING ORACLE_TIMESTAMP DUMP(ORACLE_TIMESTAMP)
11.2.0.2.0/11/2 Sun Mar 09 01:50:00 EST 2014 09-MAR-14 01.50.00.000000000 AM Typ=180 Len=7: 120,114,3,9,2,51,1
11.2.0.2.0/11/2 Sun Mar 09 03:00:00 EDT 2014 09-MAR-14 03.00.00.000000000 AM Typ=180 Len=7: 120,114,3,9,3,1,1 --Invalid timestamp
11.2.0.3.0/11/2 Sun Mar 09 01:50:00 EST 2014 09-MAR-14 01.50.00.000000000 AM Typ=180 Len=7: 120,114,3,9,2,51,1
11.2.0.3.0/11/2 Sun Mar 09 03:00:00 EDT 2014 09-MAR-14 03.00.00.000000000 AM Typ=180 Len=7: 120,114,3,9,4,1,1 --Correct timestamp
You'll notice that the 5th byte of the timestamp on the second row for 3:00 AM is incorrect (3). This was inserted using the 11.2.0.2 version. The same value inserted with the 11.2.0.3 version can be found on the fourth line with the correct 5th byte (4).
The long term fix here is to update our JDBC driver. The short term fix here was to find the rows that have the bad values and run an update statement on them from SQL Plus to set the time again (using the same value but SQL Plus will convert them correctly).
Related
Different time sysdate and current_date when connect to database since application server
I have a problem with the time that Oracle gives me when working with applications. I currently connect from an application server (apx1) to a server containing Oracle 19c (bdx1) for database operations. Both the application server and the database server are in the Canary Islands, so it must show/work local time there (GMT +1 in summer, GMT in winter). If I launch the date command on the servers, the time appears correct. [oracle#apx1 oracle]$ date vie jul 1 14:25:49 BST 2022 [oracle#bdx1 ~]$ date vie jul 1 14:25:46 BST 2022 However, after making the connection from apx1 to bdx1, I get this time offset: SQL> select systimestamp, current_timestamp, localtimestamp from dual; SYSTIMESTAMP --------------------------------------------------------------------------- CURRENT_TIMESTAMP --------------------------------------------------------------------------- LOCALTIMESTAMP --------------------------------------------------------------------------- 01/07/22 15:27:39,762772 +02:00 01/07/22 14:27:39,762775 +01:00 01/07/22 14:27:39,762775 The SESSIONTIMEZONE parameter returns the correct value, +1: SQL> SELECT SESSIONTIMEZONE FROM DUAL; SESSIONTIMEZONE --------------------------------------------------------------------------- +01:00 However, if I run this query (from the application server to the database): sac#apx1> sqlplus user/pass#bdx1 SQL> select to_char(sysdate, 'dd mm yyyy hh24:mi:ss') "SYSDATE", to_char(current_date, 'dd mm yyyy hh24:mi:ss') "CURRENT_DATE" from dual; SYSDATE CURRENT_DATE ------------------- ------------------- 01 07 2022 16:26:23 01 07 2022 15:26:23 SYSDATE result is incorrect with respect to the time that should appear.
What do you mean by "configure to correct this offset"? LOCALTIMESTAMP returns the current date and time in the session time zone in a value of datatype TIMESTAMP. CURRENT_TIMESTAMP returns the current date and time in the session time zone in a value of datatype TIMESTAMP WITH TIME ZONE. SYSTIMESTAMP returns the current date and time in the time zone of database server's operating system in a value of datatype TIMESTAMP WITH TIME ZONE. Set the SESSIONTIMEZONE and/or database server time zone properly.
The solution was posted on DBA stackexchange: https://dba.stackexchange.com/a/314026/175060. You have to set the parameter at cluster level: srvctl setenv database -db DBNAME -env "TZ=Atlantic/Canary". And then restart the DB hosts, since a CRS restart is not enough.
Change Oracle ADW default timezone
The ADW is configured for UTC timezone and there is need to change it to EST timezone. The ETL custom application has many stored custom procedures which SYSDATE returns the current date and time set for the operating system. What is the best approach to solve this problem in ADW?
To change timezone please refer here. Note that this does not change the output of SYSDATE as that relies on the OS timezone which cannot be changed. The recommendation is to use CURRENT_DATE instead. Upgrade Example This example illustrates updating DST behavior to Oracle Database 11g, release 2 for which the default time zone version is 14. First, assume that your current database is using time zone version 3, and also assume you have an existing table t, which contains timestamp with time zone data. Connect to the database as the user scott and execute the following statements: DROP TABLE t; CREATE TABLE t (c NUMBER, mark VARCHAR(25), ts TIMESTAMP WITH TIME ZONE); INSERT INTO t VALUES(1, 'not_affected', to_timestamp_tz('22-sep-2006 13:00:00 america/los_angeles', 'dd-mon-yyyy hh24:mi:ss tzr tzd')); INSERT INTO t VALUES(4, 'affected_err_exist', to_timestamp_tz('11-mar-2007 00:30:00 america/st_johns', 'dd-mon-yyyy hh24:mi:ss tzr tzd')); INSERT INTO t VALUES(6, 'affected_no_err', to_timestamp_tz('11-mar-2007 01:30:00 america/st_johns', 'dd-mon-yyyy hh24:mi:ss tzr tzd')); INSERT INTO t VALUES(14, 'affected_err_dup', to_timestamp_tz('21-sep-2006 23:30:00 egypt', 'dd-mon-yyyy hh24:mi:ss tzr tzd')); COMMIT; Then, optionally, you can start a prepare window to check the affected data and potential semantic errors where there is an overlap or non-existing time. To do this, you should start a window for preparation to migrate to time zone version 14. It is assumed that you have the necessary privileges. These privileges are controlled with the DBMS_DST package.
Determine if spring or fall and adjust time if DST or not
I have a query like this: SELECT to_date(to_char((from_tz(to_timestamp(to_char(my_column, 'YYYY-MM-DD HH:MI:SS PM'), 'YYYY-MM-DD HH:MI:SS PM') ,'America/Edmonton')at time zone 'America/Vancouver'),'YYYY-MM-DD'),'YYYY-MM-DD') as Date_Column FROM my_table It converts the time between MST to PST. This works most of the time except when daylight savings time occurs. Then a situation happens when in the clock moves forward an hour, and then since an hour is lost (for example 1:59 to the 3:00 am) then we are trying to convert a fictional time. So I am aware of the issue, and I know Oracle will throw this error for that reason: ORA-01878: specified field not found in datetime or interval I've looked all over the Internet for a possible work around but can't really find anything that works. I just want to modify my query so that it can tell what time of year it is and either do the conversation when the time exists, or handle the time somehow when it does not exists. Perhaps add the extra hour when it's missing, and remove it when it's the other time of year. Does anyone have a solution for this? Is this an impossible problem?
Instead of all the conversions you are using, just use AT TIME ZONE. You may need to take extra care depending on the data type of your column. The AT TIME ZONE functionality will always take into account daylights savings time. If your column is a TIMESTAMP WITH TIME ZONE, that would be ideal because you could just say AT TIME ZONE 'PST' and you would get the time in PST. If the column is the datatype TIMESTAMP, when you use AT TIME ZONE it will convert from the time zone of the current session to whatever you specify. If your column is a DATE, you will need to convert it to a TIMESTAMP first, then you can use the AT TIME ZONE function. Converting the DATE to TIMESTAMP will make the time zone 0:00 offset so be sure to account for that. Another handy trick is to use AT LOCAL. This will convert the timestamp to the time zone of the current session. This may be handy if you have users of your application that are in multiple time zones. Example I am in the time zone -04:00 (East coast US) which is why that appears as my time zone in the examples below. WITH my_table AS (SELECT TIMESTAMP '2021-01-01 8:00:00 -7:00' AS ts_with_tz, TIMESTAMP '2021-01-01 8:00:00' AS ts, DATE '2021-01-01' + (8 / 24) AS dt FROM DUAL) SELECT 1 AS example_num, t.ts_with_tz, t.ts, t.dt FROM my_table t UNION ALL SELECT 2 AS example_num, t.ts_with_tz AT TIME ZONE 'PST', t.ts AT TIME ZONE 'PST', TO_TIMESTAMP (t.dt) AT TIME ZONE 'PST' FROM my_table t ORDER BY 1; RESULT EXAMPLE_NUM TS_WITH_TZ TS DT ______________ ______________________________________________________ ______________________________________________________ ______________________________________________________ 1 01-JAN-21 08.00.00.000000000 AM -07:00 01-JAN-21 08.00.00.000000000 AM AMERICA/NEW_YORK 01-JAN-21 08.00.00.000000000 AM AMERICA/NEW_YORK 2 01-JAN-21 07.00.00.000000000 AM AMERICA/LOS_ANGELES 01-JAN-21 05.00.00.000000000 AM AMERICA/LOS_ANGELES 31-DEC-20 09.00.00.000000000 PM AMERICA/LOS_ANGELES
Oracle 11g converts timestamp assuming time is Cental when it already is UTC
My company recently upgraded to Oracle 11g. I have a field - Timestamp with Time Zone that has data inserted in as "UTC". We query the data, make some changes and run a generic update statement that updates all the fields in the table. Since the 11g upgrade, the update statement converts the date/time assuming it's US/Central (my time zome). Example: 7/4/2015 7:40:04.000000 AM +00:00 turns into 7/4/2015 12:40:04.000000 AM +00:00 with an update statement like this: Update Tablename set FIELD_TIMESTAMP = to_timestamp('07/04/2015 7:40:04 AM', 'mm/dd/yyyy hh:mi:ss AM') at time zone 'UTC' where ID = 201924; I have tried just inserting a basic date, I have tried using different variations of the time zone. Nothing works, it always converts the timestamp. How can I indicate to Oracle that the timestamp I'm updating is UTC and not US/Central time zone so it doesn't convert it?
Try using to_timestamp_tz. For example: SQL> create table tz_test ( id number, col1 timestamp with time zone ) Table created. SQL> insert into tz_test values (1, systimestamp) 1 row created. SQL> insert into tz_test values (2, to_timestamp('20150716 7:40:04', 'YYYYMMDD HH24:MI:SS')) 1 row created. SQL> insert into tz_test values (3, to_timestamp_tz('20150716 7:40:04 -0:00', 'YYYYMMDD HH24:MI:SS TZH:TZM')) 1 row created. SQL> commit Commit complete. SQL> select * from tz_test ID COL1 ---------- ----------------------------------- 1 7/16/2015 1:29:47.031650 PM -04:00 2 7/16/2015 7:40:04.000000 AM -04:00 3 7/16/2015 7:40:04.000000 AM +00:00 3 rows selected. My timezone is eastern (-4 from UTC), but the 3rd insert I specify the timezone to be used.
Reason behind this issue is that the TZ variable is not set properly in your.bash_profile. As of now I am almost certain you database will be returning the systimestamp in central SELECT SYSTIMESTAMP FROM DUAL; In order to permanently fix (like you had prior to upgrade to Oracle 11g) you need to add or edit TZ variable to UTC. For instance if I want to change my timezone to Asia/Calcutta then I have to add/edit the TZ variable like mentioned below TZ='Asia/Calcutta'; EXPORT TZ Once you edit the .bash_profile you need to restart the listener service and database. Or as an alternative you use can to_timestamp() and to_timestamp_tz() function as suggested by tbone to resolve this . Regards Andy
How oracle stores timestamp with time zone internally
Based on Oracle document, it stores different parts of timestamp with time zone internally in terms of numbers. I read this article http://www.orafaq.com/wiki/Timestamp which explains the algorithm of internal format of timestamp. So I did a simple test to verify it. SQL> create table tz_test(id number, tz timestamp with time zone); Table created. SQL> insert into tz_test values(1, timestamp '1999-10-29 21:00:00 -7:00'); 1 row created. SQL> insert into tz_test values(2, timestamp '1999-10-29 21:00:00 US/Pacific'); 1 row created. SQL> select id, dump(tz, 10) from tz_test where tz=timestamp '1999-10-29 21:00:00 -7:00'; ID DUMP(TZ,10) -------------------------------------------------------------------------------- 1 Typ=181 Len=13: 119,199,10,30,5,1,1,0,0,0,0,13,60 2 Typ=181 Len=13: 119,199,10,30,5,1,1,0,0,0,0,137,156 The article in orafaq talks about how oracle stores the time zone offset and my first row of the test proved that. But there's nothing about how to store the time zone literal. So I'm keen on knowing it. I also want to know internally how oracle evaluate timestamp '1999-10-29 21:00:00 -7:00' and timestamp '1999-10-29 21:00:00 US/Pacific' are identical.
They appear to be the region IDs from the time zone file, e.g. $ORACLE_HOME/oracore/zoneinfo/timezone_14.dat. In this SQL Fiddle, a value with time zone region PST8PDT has the last two bytes stood as 0x83, 0x64; which matches the values suggested by MOS document ID 414590.1. I can't see any reference that lists all the possible values, or any obvious way to examine the time zone file. It doesn't seem to be documented, which isn't unreasonable for an internal format. You could figure out values you're interested in, or all values, using the same method but I'm not sure why you'd want to. There's another support document you might find useful, 340512.1, which is a FAQ about timestamps and time zones. To convert a region name to an offset value you can use the tz_offs function, but it uses the current date - so the answer you get will depend on whether you ask in the winter or summer. To get the offset for a certain date you need to compare that timestamp with a UTC timestamp. And you can't really go the other way, as multiple time zone regions map the same offset.
TIMESTAMP '1999-10-29 21:00:00 -7:00' and TIMESTAMP'1999-10-29 21:00:00 US/Pacific' are not identical. You see the difference when you run this query for example: SELECT TIMESTAMP '1999-12-29 21:00:00 -7:00', TIMESTAMP '1999-12-29 21:00:00 US/Pacific', TIMESTAMP '1999-06-29 21:00:00 -7:00', TIMESTAMP '1999-06-29 21:00:00 US/Pacific' FROM dual; 29.12.1999 21:00:00.000000000 -07:00 29.12.1999 21:00:00.000000000 -08:00 29.06.1999 21:00:00.000000000 -07:00 29.06.1999 21:00:00.000000000 -07:00 US/Pacific considers Daylight Saving Times, -7:00 does not!