From EGIWiki
Revision as of 17:57, 27 January 2016 by Vbrillau (talk | contribs) (Resource Center Responsibilities)
Jump to: navigation, search
Main operations services Support Documentation Tools Activities Performance Technology Catch-all Services Resource Allocation Security

Documentation menu: Home Manuals Procedures Training Other Contact For: VO managers Administrators

Baustelle.png This page is under construction.
Title EGI CSIRTSecurity Incident Handling Procedure
Document link
Last modified V5
Policy Group Acronym EGI-CSIRT
Policy Group Name EGI-CSIRT
Contact Group
Document Status DRAFT
Approved Date
Procedure Statement
Owner Owner of procedure


This procedure is aimed at minimising the impact of security incidents by encouraging post-mortem analysis and promoting cooperation between Resource Centers.
It is based on the Security Incident Response Policy.

This incident response procedure is aiming at complementing local security procedures. Unless specified otherwise in separate service level agreements, all times in this document refer to normal local working hours.

This document is intended for Resource Center security contacts and administrators and is primarily aimed at reporting, investigating and resolving security incidents.


Please refer to the EGI Glossary for the definitions of the terms used in this procedure.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", “MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119.

Contact points

  • abuse at Address to be used for reporting security Incident (In case of TLP:RED data, use GPG: A97F 3BDD F0EE 01A1 176C C13A 93BF 7F91 5696 F750)
  • site-security-contacts at Mailing list containing all Resource Center "CSIRT E-Mail" as defined in GOC-DB

Entities involved in the procedure

  • EGI-CSIRT Security Officer on Duty: irtf at
  • NGI Security Officer: NGI Security E-Mail as defined in GOC-DB
  • Resource Center: RC CSIRT E-Mail as defined in GOC-DB

Resource Center Responsibilities

The following table describes the actions to be taken when an incident potentially affecting EGI users, data, services, infrastructure is suspected. Administrators are recommended to take note of every action (with timestamp) they take, for later analysis or legal cases.

Step Action Deadline
1 Inform your local security team, your NGI Security Officer and the EGI CSIRT via You are encouraged to use the recommended templates. Within 4 hours of discovery
2 Isolate the incident while keeping all forensic data: Do NOT reboot or power off hosts. Do NOT destroy VMs. Instead disconnect it from the network and if possible take a snapshot of the system. In case you need support, contact your local security team or the EGI CSIRT via Within 1 day of discovery
3 Together with your local security team and the EGI CSIRT decide if it is an incident
4 If applicable, announce downtime for the affected services in accordance with the EGI Operational Procedures Within 1 day of isolation
5 Perform appropriate analysis and take necessary corrective actions, see Incident Analysis Guideline within 4 hours of any EGI CSIRT request
6 Coordinate with your local security team and the EGI CSIRT to send an incident closure report within 1 month following the incident to the EGI CSIRT via, including lessons learnt and resolution. This report should be labelled AMBER or higher, according to the Traffic Light Protocol. within 1 month of incident resolution
7 Restore the service and, if needed, update the service documentation and procedures to prevent recurrence as necessary.

EGI-CSIRT Responsabilities

EGI-CSIRT Security Officer on Duty

The EGI-CSIRT Security Officer on Duty tasks are:

  • Evaluate the initial incident report and determine whether it appears to be part of an incident covering multiple RCs, in particular, whether it is related to a previously known incident (e.g. do the same attacking IP addresses appear, are the attacker's tools and methodology strongly similar):
    • If this is a new, unrelated incident, assign an identifying tag (of the format [EGI-YYYYMMDD-NN]) to the incident and announce it to all RCs via using the recommended templates.
    • If the incident is part of an incident covering multiple RCs, the incident coordinator MAY choose not to announce each incident separately, but instead issue regular updates on the overall incident.
  • Whenever and as often as necessary, send updated:
    • Summary reports to all RCs (, containing the status of the incident and indicators of compromise that can be used by RCs to evaluate their implication
    • detailed reports to the RCs directly involved and affected by the incident, containing interesting findings or possible leads that could be used to resolve the incident
  • Whenever an obviously malicious behaviour or policy violation can be linked to a user account or identity:
    • Add the account or identity to the emergency suspension list following the appropriate procedure.
    • If applicable, report the incident to the VO providing access
    • Verify the legimity of the activity with the owner of the account or identity
  • When suspended accounts or identities no longer represent a threat, typically when the incident is resolved and compromised credentials have been re-issued, remove them from the emergency suspension list
  • When a virtual appliance is detected vulnerable or malicious, ensure that:
    • Its endorsement in APP-DB is revoked
    • All instantiated and running VMs using this virtual appliance are properly handled
  • Based on the incident closure report received from the affected RC, send a closure report with the relevant information to all partners.
  • Take any appropriate actions in order to:
    • Actively stimulate and probe the affected parties to obtain accurate information at an appropriate level of detail and in a timely manner.
    • Understand the exact cause and extent of the incident, what assets have been compromised (credentials etc.), and how to resolve the incident.
    • Help involved RCs resolve the incident by providing recommendations, promoting collaboration with other RCs and periodically checking their status.
    • Maintain communications with any other involved parties inside and outside EGI.

EGI-CSIRT Security Incident Coordinator

In addition, the EGI CSIRT appoints a security incident coordinator for each incident, responsible for:

  • Ensuring that the incident does not become stale due to the duty rotation.
  • When the incident is closed, based on the RC incident closure report and in coordination with the on-duty security officers, conduct a debriefing

Incident Analysis Guideline

As part of the security incident resolution process, RCs are expected to produce the following information:

  • Who/how detected or reported the incident
  • Host(s) affected (ex: compromised hosts, hosts running suspicious user code)
  • If applicable, the virtual appliance used to instantiate the affected virtual machine.
  • Possibly affected X509 certificate DNs of the user(s), operator(s), consumer(s)
  • If relevant, host(s) used as a local entry point to the RC (ex: UI or WMS IP address)
  • If available, Remote IP address(es) of the attacker
  • Evidence of the compromise, including timestamps (ex: suspicious files,log entry or network activity)
  • If available, what was lost, details of the attack (ex: compromised credentials, (root) compromised host)
  • If available and relevant, the list of other RCs possibly affected
  • If available and relevant, possible vulnerabilities exploited by the attacker
  • The actions taken to resolve the incident
  • Identify and kill suspicious process(es) as appropriate, but aim at preserving the information they could have generated, both in memory and on disk by dumping them beforehand, see Forensic_Howto
  • If it is suspected that some grid credentials have been abused or compromised, you MUST ensure the relevant accounts have been suspended centrally according to the EGI CSIRT Operational Procedure for Compromised Certificates and Central Security Emergency suspension. Contact the EGI CSIRT for this.
  • If it is suspected that some grid credentials have been abused, you MUST ensure that the relevant VO manager(s) have been informed. VO contacts are available from: Contact the EGI CSIRT for this.
  • If it is suspected that some grid credentials have been compromised, you MUST ensure that the relevant CA has been informed. CA contacts are available from: Contact the EGI CSIRT for this.
  • If it is suspected that a virtual appliance used to instantiate the affected virtual machine, you MUST check all other running instances based on the same virtual appliance and ensure that the virtual appliance is centrally de-endorsed. Contact the EGI CSIRT for the latter.
  • If needed, seek help from your local security team, from your NGI Security Officer or from the EGI CSIRT
  • If relevant, additional reports containing suspicious patterns, IP addresses, files or evidence that may be of use to other infrastructure parties SHOULD be sent to the EGI CSIRT.

As part of the investigations, RCs MUST be able to provide the relevant logging information produced by local services. Logging information such as IP addresses, timestamps and identities involved etc., concerning the source of any suspicious successful connection, must meet the following minimal requirements, if possible according to local laws:

  • 6 months prior to the discovery of the incident for successful SSH connections against EGI services, and for the originating submission host for grid jobs or virtual machines
  • 3 months prior to the discovery of the incident for all other EGI related services.

For example, should an incident be detected and reported on 1st of September, it is expected that RCs can produce the relevant logging information for suspicious SSH connections from 1st of March.

Revision History

Version Authors Date Comments
1 Mingchao Ma/STFC 28/06/2011 First Draft based on MS405, a few minor update, added appendix C and appendix D
2 Mingchao Ma/STFC 28/07/2011 Addressed comments from Dorine
3 Mingchao Ma/STFC 11/10/2011 Corrected a reference in the incident response check list (appendix D)
4 Sven Gabriel/Nikhef/FOM 26/05/2015 Adjustments for IR in EGI Cloud environments
5 Vincent Brillault/CERN Ongoing Adapted to the wiki format, several adjustment