Return-Path: X-Original-To: apmail-hadoop-hdfs-user-archive@minotaur.apache.org Delivered-To: apmail-hadoop-hdfs-user-archive@minotaur.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 3AC9318F78 for ; Mon, 8 Jun 2015 23:04:51 +0000 (UTC) Received: (qmail 39613 invoked by uid 500); 8 Jun 2015 23:04:46 -0000 Delivered-To: apmail-hadoop-hdfs-user-archive@hadoop.apache.org Received: (qmail 39481 invoked by uid 500); 8 Jun 2015 23:04:45 -0000 Mailing-List: contact user-help@hadoop.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: user@hadoop.apache.org Delivered-To: mailing list user@hadoop.apache.org Received: (qmail 39471 invoked by uid 99); 8 Jun 2015 23:04:45 -0000 Received: from Unknown (HELO spamd4-us-west.apache.org) (209.188.14.142) by apache.org (qpsmtpd/0.29) with ESMTP; Mon, 08 Jun 2015 23:04:45 +0000 Received: from localhost (localhost [127.0.0.1]) by spamd4-us-west.apache.org (ASF Mail Server at spamd4-us-west.apache.org) with ESMTP id 3B5E5C095F for ; Mon, 8 Jun 2015 23:04:45 +0000 (UTC) X-Virus-Scanned: Debian amavisd-new at spamd4-us-west.apache.org X-Spam-Flag: NO X-Spam-Score: 3.97 X-Spam-Level: *** X-Spam-Status: No, score=3.97 tagged_above=-999 required=6.31 tests=[HTML_MESSAGE=3, KAM_LAZY_DOMAIN_SECURITY=1, RCVD_IN_MSPIKE_H4=-0.01, RCVD_IN_MSPIKE_WL=-0.01, T_RP_MATCHES_RCVD=-0.01] autolearn=disabled Received: from mx1-us-west.apache.org ([10.40.0.8]) by localhost (spamd4-us-west.apache.org [10.40.0.11]) (amavisd-new, port 10024) with ESMTP id 3txczPYF537V for ; Mon, 8 Jun 2015 23:04:36 +0000 (UTC) Received: from dcwxppheip003.imr.gm.com (dcwxppheip003.imr.gm.com [198.208.69.241]) by mx1-us-west.apache.org (ASF Mail Server at mx1-us-west.apache.org) with ESMTPS id EB8FB275E9 for ; Mon, 8 Jun 2015 23:04:35 +0000 (UTC) X-SKIP-ALL: FALSE X-LoopCount2: from 148.93.0.28 Received: from dcwipvmdlp019.edc.nam.gm.com (HELO dcwxppheip003-s.imr.gm.com) ([148.93.0.28]) by dcwxppheip003-s.imr.gm.com with ESMTP; 08 Jun 2015 19:04:29 -0400 X-SKIP-ALL: FALSE X-Loop1: 1 X-IronPort-AV: E=Sophos;i="5.13,576,1427774400"; d="scan'208,217";a="293026549" Received: from dcwippexch002.edc.nam.gm.com (HELO DCWIPPEXCH002.nam.corp.gm.com) ([148.93.30.12]) by dcwxppheip003-s.imr.gm.com with ESMTP/TLS/AES256-SHA; 08 Jun 2015 19:04:28 -0400 Received: from DCWIPPEXCH006.nam.corp.gm.com (148.93.30.16) by DCWIPPEXCH002.nam.corp.gm.com (148.93.30.12) with Microsoft SMTP Server (TLS) id 15.0.1044.25; Mon, 8 Jun 2015 19:04:28 -0400 Received: from DCWIPPEXCH006.nam.corp.gm.com ([169.254.9.41]) by DCWIPPEXCH006.nam.corp.gm.com ([169.254.9.145]) with mapi id 15.00.1044.021; Mon, 8 Jun 2015 19:04:28 -0400 From: Robert B Hamilton To: "user@hadoop.apache.org" Subject: Flume rollback during restart possible? Thread-Topic: Flume rollback during restart possible? Thread-Index: AdCiPihBY5OSJHigQEKqQqAz6W8OwA== Date: Mon, 8 Jun 2015 23:04:27 +0000 Message-ID: <6c940c200b8f475aa38e7d175bfa7995@DCWIPPEXCH006.nam.corp.gm.com> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-ms-exchange-transport-fromentityheader: Hosted x-originating-ip: [10.121.8.249] Content-Type: multipart/alternative; boundary="_000_6c940c200b8f475aa38e7d175bfa7995DCWIPPEXCH006namcorpgmc_" MIME-Version: 1.0 X-CFilter-Loop: Reflected --_000_6c940c200b8f475aa38e7d175bfa7995DCWIPPEXCH006namcorpgmc_ Content-Type: text/plain; charset="utf-7" Content-Transfer-Encoding: quoted-printable Hello all. I have an interesting case where we lose data in the event of a = flume crash, which is easily reproducible when we kill -9 the flume agent. I believe that this may be because the Flume Sink is issuing a commit befo= re it actually completes the fs sync. If this is the case then the last fe= w commits just before the crash would have removed events from the queue ev= en though those events will needed to perform a recovery. My question is, = are those events still possibly in the WAL? If so, is it possible so someho= w roll back the queue to a point in time before the commits were processed,= and restart from that state? How would I accomplish this? Nothing in this message is intended to constitute an electronic signature u= nless a specific statement to the contrary is included in this message. Confidentiality Note: This message is intended only for the person or entit= y to which it is addressed. It may contain confidential and/or privileged m= aterial. Any review, transmission, dissemination or other use, or taking of= any action in reliance upon this message by persons or entities other than= the intended recipient is prohibited and may be unlawful. If you received = this message in error, please contact the sender and delete it from your co= mputer. --_000_6c940c200b8f475aa38e7d175bfa7995DCWIPPEXCH006namcorpgmc_ Content-Type: text/html; charset="utf-7" Content-Transfer-Encoding: quoted-printable +ADw-html+AD4- +ADw-head+AD4- +ADw-meta name+AD0AIg-Generator+ACI- content+AD0AIg-Microsoft Word 15 (filt= ered medium)+ACIAPg- +ADw-style+AD4- +ADwAIQ--- /+ACo- Font Definitions +ACo-/ +AEA-font-face +AHs-font-family:+ACI-MS Mincho+ACIAOw- panose-1:2 2 6 9 4 2 5 8 3 4+ADsAfQ- +AEA-font-face +AHs-font-family:+ACI-Cambria Math+ACIAOw- panose-1:2 4 5 3 5 4 6 3 2 4+ADsAfQ- +AEA-font-face +AHs-font-family:Calibri+ADs- panose-1:2 15 5 2 2 2 4 3 2 4+ADsAfQ- +AEA-font-face +AHs-font-family:+ACIAXABA-MS Mincho+ACIAOw- panose-1:2 2 6 9 4 2 5 8 3 4+ADsAfQ- /+ACo- Style Definitions +ACo-/ p.MsoNormal, li.MsoNormal, div.MsoNormal +AHs-margin:0in+ADs- margin-bottom:.0001pt+ADs- font-size:11.0pt+ADs- font-family:+ACI-Calibri+ACI-,sans-serif+ADsAfQ- a:link, span.MsoHyperlink +AHs-mso-style-priority:99+ADs- color:+ACM-0563C1+ADs- text-decoration:underline+ADsAfQ- a:visited, span.MsoHyperlinkFollowed +AHs-mso-style-priority:99+ADs- color:+ACM-954F72+ADs- text-decoration:underline+ADsAfQ- span.EmailStyle17 +AHs-mso-style-type:personal-compose+ADs- font-family:+ACI-Calibri+ACI-,sans-serif+ADs- color:windowtext+ADsAfQ- .MsoChpDefault +AHs-mso-style-type:export-only+ADs- font-family:+ACI-Calibri+ACI-,sans-serif+ADsAfQ- +AEA-page WordSection1 +AHs-size:8.5in 11.0in+ADs- margin:1.0in 1.0in 1.0in 1.0in+ADsAfQ- div.WordSection1 +AHs-page:WordSection1+ADsAfQ- --+AD4- +ADw-/style+AD4APAAh---+AFs-if gte mso 9+AF0APgA8-xml+AD4- +ADw-o:shapedefaults v:ext+AD0AIg-edit+ACI- spidmax+AD0AIg-1026+ACI- /+AD4= - +ADw-/xml+AD4APAAhAFs-endif+AF0---+AD4APAAh---+AFs-if gte mso 9+AF0APgA8-xm= l+AD4- +ADw-o:shapelayout v:ext+AD0AIg-edit+ACIAPg- +ADw-o:idmap v:ext+AD0AIg-edit+ACI- data+AD0AIg-1+ACI- /+AD4- +ADw-/o:shapelayout+AD4APA-/xml+AD4APAAhAFs-endif+AF0---+AD4- +ADw-/head+AD4- +ADw-body lang+AD0AIg-EN-US+ACI- link+AD0AIgAj-0563C1+ACI- vlink+AD0AIgAj-9= 54F72+ACIAPg- +ADw-div class+AD0AIg-WordSection1+ACIAPg- +ADw-p class+AD0AIg-MsoNormal+ACIAPg-Hello all. I have an interesting case = where we lose data in the event of a flume crash, which is easily reproduci= ble when we kill -9 +ACY-nbsp+ADs-the flume agent.+ADw-o:p+AD4APA-/o:p+AD4A= PA-/p+AD4- +ADw-p class+AD0AIg-MsoNormal+ACIAPgA8-o:p+AD4AJg-nbsp+ADsAPA-/o:p+AD4APA-/= p+AD4- +ADw-p class+AD0AIg-MsoNormal+ACIAPgAm-nbsp+ADs-I believe that this may be = because the Flume Sink is issuing a commit before it actually completes the= fs sync. +ACY-nbsp+ADs-If this is the case then the last few commits just = before the crash would have removed events from the queue even though those events will needed to perform a recovery.+ACY-nbsp+ADs- My question is, ar= e those events still possibly in the WAL? If so, is it possible so somehow = roll back the queue to a point in time before the commits were processed, a= nd restart from that state? How would I accomplish this?+ADw-o:p+AD4APA-/o:p+AD4APA-/p+AD4- +ADw-p class+AD0AIg-MsoNormal+ACIAPgA8-o:p+AD4AJg-nbsp+ADsAPA-/o:p+AD4APA-/= p+AD4- +ADw-p class+AD0AIg-MsoNormal+ACIAPgA8-o:p+AD4AJg-nbsp+ADsAPA-/o:p+AD4APA-/= p+AD4- +ADw-p class+AD0AIg-MsoNormal+ACIAPgA8-o:p+AD4AJg-nbsp+ADsAPA-/o:p+AD4APA-/= p+AD4- +ADw-p class+AD0AIg-MsoNormal+ACIAPgA8-o:p+AD4AJg-nbsp+ADsAPA-/o:p+AD4APA-/= p+AD4- +ADw-/div+AD4- +ADw-br+AD4- +ADw-br+AD4- Nothing in this message is intended to constitute an electronic signature u= nless a specific statement to the contrary is included in this message. +ADw-br+AD4- +ADw-br+AD4- Confidentiality Note: This message is intended only for the person or entit= y to which it is addressed. It may contain confidential and/or privileged m= aterial. Any review, transmission, dissemination or other use, or taking of= any action in reliance upon this message by persons or entities other than the intended recipient is prohib= ited and may be unlawful. If you received this message in error, please con= tact the sender and delete it from your computer. +ADw-/body+AD4- +ADw-/html+AD4- --_000_6c940c200b8f475aa38e7d175bfa7995DCWIPPEXCH006namcorpgmc_--